Thread by @lastpositivist

Ok! Day 2 of Rethinking Formal Methods in Philosophy #RFMP2019 - today we're talking metaphilosophy, and some concrete examples of formal tools in philosophy. We're starting with a talk by Branden Fitelson (@fitelson) How to Model Epistemic Probabilities of Conditionals

https://twitter.com/lastpositivist/status/1175056066595446787

Literal galaxy brain take? Not so, #RFMP2019 host Samuel Fletcher assures us - that’s just what Massachusetts looks like.

His paper is on Lewis work on Probabilities of Conditionals: Fitelson says in early work Lewis made a subtle mistake about what how formal methods can fruitfully interact with philosophy, and so misled a generation into rejecting Adam's Thesis (philsci-archive.pitt.edu/15208/1/gigo8c…)

Lewis 1976 paper Probability of Conditionals and Conditional Probabilities involved a strong assumption that all rational requirements hold in a very strong sense but his 1980 paper Subjectivist's Guide to Objective Chance involved a weaker requirement. The latter is preferable.

As Fitelson notes in his recent paper (onlinelibrary.wiley.com/doi/full/10.10…) Lewis implicitly required that if Pr(P -> Q) = Pr (Q | P) is rationally required, then this equation must hold no matter what you learn ends. This ends up being far too strong, Fitelson holds.

"Lewis says some stuff which seems to support his claim - but of course it doesn't, because it's false." - Fitelson #RFMP2019

Fitelson shows (as Lewis himself knew) if you require the Principal Principle (fitelson.org/coherence/pett…) to hold fully resiliently, in the same way that the conditional probability equation was implicitly meant to hold, then an exactly equivalent argument reduces it to triviality.

Rather than take it as a reductio of the Principal Principle, Lewis introduces a notion of admissibility that rules out certain things one might learn as really bearing upon the claim made by the Principal Principle. Fitelson thinks the same move should be made for Adam's thesis.

New version of Adam's thesis equation should be:

For all factual propositions, p, q, and x:

Pr( p -> q | x) = Pr (q | p & x) only if x satisfies the following condition: screening off - Pr (q | x & p) = Pr (q | p)

[The p's and q's are not capital letters now. Deal with it.]

"Intuitively, when we advise a certain level of confidence in p -> q, we are presupposing that the advisee doesn't know anything which trumps the informational connection between p and q". - Fitelson #RFMP2019

Fitelson: counter-examples to the claim that Pr(p->q) = Pr(q|p) involve learning some x that is inadmissible according to screening off condition. This is a standing challenge to the field, and welcomes people sending him counter-examples - fitelson.org #RFMP2019

Fitelson gives shout outs to work by Bradley users.ox.ac.uk/~sfop0776/Brad… and Hajek pdfs.semanticscholar.org/b08d/7678968bd… - but in general people were too quick to reject Adam's thesis and so debate was held back. We should have been charitable, we were overly impressed by Lewis' technical result.

Question from the audience: what's meant by rationality in this literature? Fitelson notes folk often take on board probabilism for economic-instrumental-rationality-esque reasons plato.stanford.edu/entries/episte… But more is needed and he thinks formal results won't settle this. #RFMP2019

Editorialising a bit, I think Fitelson's moral is about limits of formal results relevance to philosophical questions. Formalism lets us state clearly our assumptions and study their interrelations; this is very useful. But can't tell us what to accept in the 1st place! #RFMP2019

Some on the spot formal philosophy here at #RFMP2019

Next stage is reasoning about nested conditionals and modal reasoning. The work never ends for the formal philosopher! #RFMP2019

"We should let a thousand conditional flowers bloom!" - Sam Fletcher making case for pluralistic semantics of conditionals, and Fitelson agrees. He just wants to see the various explications given an explication. Lewis triviality results placed a roadblock in the way of inquiry.

Question from audience about what is meant by "factual claims", Fitelson clarifies that Adams meant something you can bet on where you know how to settle the bet, Fitelson himself generally means direct claims about the actual world, so no nested conditionals or modal claims.

Question from audience - what are priors? [All the Bayesians excuse themselves from the room, look at their shoes, say "what in the world is that behind you!", etc] #RFMP2019

(Fitelson's answer is that priors are a hypothetical construct, useful as a modelling tool but not appropriately interpreted as real features of actual agents.)

"I'm not a uniqueness person, I don't think your evidence requires a unique credence" - Fitelson on his own views. In general discussion led us into subtleties of the deontic assumptions being made in arguments about rational norms, clear that fine subtleties make big difference.

Ok next up we have Cailin O’Connor (@cailinmeister): "Synthesizing Formal and Empirical Methods in Philosophy: Testing the Red King" - let's go #RFMP2019

@cailinmeister Cailin says her talk is gonna have three layers - an inner point which is her first order thesis, a middle level metathesis, and an outer most overall metaphilosophical reflection

And each layer gets an introductory slide! #RFMP2019

Cailin is going to be drawing from her work on the Cultural Red King Effect, readers can check out more of her work on this here.

tandfonline.com/doi/full/10.10…

O’Connor is building on biological work on the Red Queen effect, and using simple game theoretic models to try and capture key related features of the Red King effect. Already this is deeply interdisciplinary work. #RFMP2019

(@CT_Bergstrom just got a shout out for his work with Lachmann, see pnas.org/content/100/2/…)

@CT_Bergstrom Sometimes in Nash demand games evolving more slowly helps you end up in an equilibrium you prefer; in the short run it is beneficial to make low demands, which do well against all comers. But once fast evolving side has moved towards low demands you can move in and demand higher.

O’Connor notes that Bruner first applied this to the cultural case, referencing this paper academia.edu/29685192/Minor…

... and then O’Connor and other collaborators (including yours truly!) explored how the red king hypothesis plays out in a variety of circumstances, here’s our work liamkofibright.com/uploads/4/8/9/… - O’Connor also notes the connection to Du Boisian ideas about double consciousness.

[This talk is taking us through the process of developing a research project in formal philosophy, very instructive for folk interested in how the sausage is actually made #RFMP2019]

O'Connor made sure to generalise the Red King hypothesis to different types of models just to ensure the results were robust, and also to look at models less directly taken from biology and more specifically attuned to looking at cultural phenomena. She found results held up.

Exploring models also allowed her to look at which conditions led to a Red King versus Red Queen effect. She found making agents risk averse further disadvantages minority groups, with stronger red king effects. Relates to literature saying poorer people are more risk averse.

Adding in group preference had a similar effect in the models she looked at - it turns out making models more realistic tended to predict greater minority group disadvantage (should have mentioned that it was minority groups in the models who are subject to Red King disadvantage)

But of course, these are all still just models, and inevitably highly simplified models! Why think it applies to the real world? O'Connor said this led to natural next question: could minority status alone generate bargaining disadvantage with real humans in experiments?

O'Connor keen to stress that it was only because of the formal work that they even thought to ask the empirical question, it's not a priori obvious that *sheer numerical difference in group size* leads to this particular type of disadvantage. #RFMP2019

So on to experimental work, bargaining in the lab! O’Connor mentions that they had the experimental design idea because Hannah Rubin was taking an experimental economics class at the time, further interdisciplinary cross fertilisation. Here’s the paper: researchgate.net/publication/33…

[O'Connor mentions that Aydin Mohseni is still a grad student and will be on the market soon, and if anyone wants to hire him he's great! aydinmohseni.com]

Experimental results supported conclusion! They'd love to do bigger and more powerful tests, but its expensive and time consuming. In any case, formal and empirical results pointing to similar conclusion about real world dynamics of minority disadvantage. #RFMP2019

Re middle layer, what does this tell us? Simple models cannot by themselves tell us much about the real world. But they can give us how-possible knowledge, counter-factual knowledge, be used as aids to thought experiments, and direct empirical research. The latter happened here.

Re outer layer, began with biological modelling methods, applied them to quite different problems in feminist social theory, then complemented this with methods from behavioural econ. Opportunistic adoption of multi-method interdisciplinary work very fruitful. #RFMP2019

O'Connor mentions that there is some work in the history and sociology of science which suggests that a lot of scientific breakthroughs happen when people simply bring together methods and questions from different fields. Why not apply this to philosophy too?

Though O'Connor tempers this with a note saying that fields probably need some glue that holds them together and facilitates shared conversations, and this requires some methodological commonality; relevant to #RFMP2019 she thinks some logic should be part of that in philosophy.

"I just think we should get a bit weird with methodology" - O'Connor (@cailinmeister) #RFMP2019

@cailinmeister In response to question O'Connor mentions that what makes this possible is allowing grad students to go and do training in totally different fields, idiosyncratically and based on the particular questions they are interested in. Highly recommends this interdisciplinary approach.

@cailinmeister Ok! Next up we have Catarina Dutilh Novaes (@cdutilhnovaes) with her paper Carnapian Explication and Ameliorative Analysis: A Systematic Comparison. We all knew we'd get to Carnap eventually here at the #RFMP2019

@cailinmeister @cdutilhnovaes (Editorialising: the introduction by Roy Cook is sweet, obviously old friends with Catarina, mentions her great work on medieval logic, thinking about what logic is and what it should do for philosophy, and feminist philosophy and professional cultural change #RFMP2019)

@cailinmeister @cdutilhnovaes There's also a title change! @cdutilhnovaes 's talk is actually called "Carnapian explication as an ameliorative project: logic and social change".

@cailinmeister @cdutilhnovaes Begins by noting Carnapian explication has made a come back as a methodological framework for the use of formal methods in philosophy, where as recently as 10 years ago Carnap was viewed as totally outmoded logical positivist, now increasingly seen as central to the discipline.

In a similar time frame there has been a growth in interest in the relationship between philosophy and social change, seeing how one can be a positive tool for another - shout out to @jasonintrator who she says really influenced this trend and her thinking in particular

So Catarina is gonna argue that these two trends of more socially relevant philosophy and more Carnapian explication can complement each other! #RFMP2019

Catarina is gonna be drawing from some of her previously published work here, so check out: link.springer.com/article/10.100… and link.springer.com/article/10.100… - 2nd paper ended up generating a lot of discussion on Reddit: "don't underestimate the power of reddit", she really liked the thread.

She also recommends this book! So see if you can check it out (and find a cheaper version than linked here, yikes!)

cambridge.org/us/academic/su…

"The task of explication consists in transforming a more or less inexact concept into an exact one, or, rather, in replacing the first by the second." - Carnap, Catarina notes degree of exactness is comparative not absolute and the idea is we can iterate this replacement process.

Four desiderata for an explication according to Carnap:

- simplicity
- exactness
- similarity
- fruitfulness

Catarina begins by discussing whether the exactness desiderata compels us to scientism? #RFMP2019

Important to bear in mind, Catarina says, fruitfulness is ultimately the goal, exactness is only justified in so far as it is pragmatically useful in helping us seecuring this itself pragmatic goal. He doesn't, and doesn't intend for us, to fetishise exactness for its own sake.

What about fruitfulness itself? (We're taking simplicity as a project for other philosophy of science!) Well Carnap doesn't actually say much. He says that being "useful for the formulation of many universal statements", reminiscent of Fregean notion of useful definitions

(Oh also "similarity" is meant to just help us keep our analysis on target, ensuring we're talking about broadly the same thing. Not that important, Catarina says.)

Catarina quotes from one of Carnap’s famous examples of a fruitful explication. She notes that it’s not mathematical, that’s not a requirement of being a good explication, and clearly part of what’s making the explication fruitful is facilitating systematisation.

Quoting Carus, Catarina summarises Carnap's Enlightenmentism "the ambition of shaping individual and social development on the basis of better and more reliable knowledge than the tangled, confused, half-articulate but deeply rooted conceptual system inherited from our ancestors"

She thinks that this project is guiding Carnap's projects throughout his life, and can be seen as in the background of his ideas about what would constitute fruitful explication.

This seems to match Jeffrey’s, Carnap’s student, interpretation of Carnap’s project. Choosing our formalism is an expression of our autonomy, an application of the enlightenment ideal that we dare to think for ourselves! #RFMP2019

So how does this connect to the claim about seeking universal statements? Such statements allow us to gain "rational controls over hypothetical plans of action", they help us make rational social decisions about what policies to implement or institutions to construct.

We move on to discussing Sally Haslanger’s work (maybe check out onlinelibrary.wiley.com/doi/abs/10.111… as pertinent here)

Haslanger's ameliorative analysis: "ameliorative projects, in contrast, begin by asking: what's point of having concept in question; for example, why do we have a concept of knowledge or a concept of belief? What concept would work best?...

...In limit case a theoretical concept is introduced by stipulating the meaning of a new term, and its content is determined entirely by the role it plays in the theory." - Catarina notes limit case sounds just like Carnappian explication, that's the connection she wants to draw.

She illustrates how Carnapian explication and Haslangerian ameliorative analysis can be done in the same vein, at the same time, and to the same end, by looking at a paper on intersectionality theory. Second shout out to @MorganKThomp at the #RFMP2019!

liamkofibright.com/uploads/4/8/9/…

@MorganKThomp "Formal approaches not only assist clarity and precision, but also afford opportunities for aiding social engineering... of course, one can always engineer things to go in the wrong direction too, but I am trying to keep things optimistic today" - Catarina Dutilh Novaes #RFMP2019

@MorganKThomp Questioner worries fruitfulness can always be fruitful for bad things, don't we need to real facts about the good and only allow fruitful towards that? Catarina agrees. Next paper will be on using genealogical methods to supplement explication, to better identify good projects.

@MorganKThomp Ok we're back from lunch, and it's @Prof_Livengood talking about "Second Thoughts: On Logic and Formal Methods in Philosophy" #RFMP2019

@MorganKThomp @Prof_Livengood Prof. Livengood begins by thinking of the questions: how much mathematics is required to do good philosophical work? Should we admit only mathematicians to philosophy programmes? (Plato's "none but geometers enter here" slogan) What do I want my own students to know?

@MorganKThomp @Prof_Livengood He's going to argue that we should get rid of all course requirements, no particular required knowledge for philosophy students to come away with! Getting spicy up in the #RFMP2019 !

@MorganKThomp @Prof_Livengood "I think mathematics is beautiful and powerful. But I also find it enormously difficult and frustrating. I'm not very good at it. And although I wish I were good at it, I haven't practiced it nearly enough to become good at it" - Livengood begins by keeping it real. #RFMP2019

@MorganKThomp @Prof_Livengood Livengood described his background as an electrical engineer; towards the end of his degree he took philosophy courses and they "totally wrecked his world", undermining convictions about religion and how science worked, and inducing great doubt about all aspects of his worldview.

@MorganKThomp @Prof_Livengood One result of this was that he was converted to how Peirce thinks about logic, a pragmatist conviction he retains to this day - and which puts him badly out of line with how philosophers think about logic. #RFMP2019

@MorganKThomp @Prof_Livengood "I went to grad school still thinking I would be a logician. Because -- and I really can't emphasise this enough - I was dumb." @Prof_Livengood getting some subtle meme magic into his talk. #RFMP2019

The theses under discussion here #RFMP2019

(This blurry picture only hints at the excellent tie game that @Prof_Livengood has brought to this conference.)

@Prof_Livengood "Logic... may be considered as the Science and also as the Art, of Reasoning. It investigates the principles on which argumentation is conducted, and furnishes tools to protect the mind from error" - Richard Whately (I might have slightly missed the quote here)

"Logic is not the science of Belief, but the science of Proof, or Evidence. In so far as belief professes to be founded on proof, the office of logic is to supply a test for ascertaining whether or not the belief is well grounded" - J.S. Mill #RFMP2019

"It is impossible to direct attention too prominently to the fact that logic (and therefore Probability as a branch of logic) is not concerned with what men do believe, but with what they ought to believe, if they are to believe correctly" - John Venn

@Prof_Livengood taking us on a tour through thinkers who share his conception of logic.

@Prof_Livengood Livengood says the main villains end up being Russell and Quine in giving people a picture of logic that he takes to be mistaken and unhelpful.

@Prof_Livengood General theme emerging from Livengood's talk so far is that a proper conception of logic as a normative science of good reasoning is that probabilistic or inductive reasoning ends up just as much a proper part of logic as what we now think of as logic-proper, deductive logic.

@Prof_Livengood I didn't get the quotes in time but I want everyone to know that Wittgenstein just got rekt, so take it as given that you should be super angry all yinz fanfolk out there should be super angry on the internet because your bot got owned. It... it's getting late. #RFMP2019

Livengood quotes Priest: "We all reason. We try to figure out what is so, reasoning on the basis of what is so. We try to persuade others of our opinion, on the basis of reasons. Logic is the study of what counts as a good reason for what, and why." He likes this vision of logic.

A (non deductive!) argument for logic as the normative science of reason.

With this broad vision of logic in mind Livengood argues that people should have options available to them with plenty of logic-broadly-construed available as options, yet there's no one thing uniquely required to reason well philosophically, and so no one required course.

A cheeky formalisation - Livengood thinks that this will speak against any requirement in philosophy .

Another argument Livengood gives - requirements place roadblocks to student engagement, where people struggle with being made to master any one method even though they are quite competent at applying the reasoning tools they need for their project. Why force them to do logic?

A caveat: this depends on the particular nature of the course you are teaching, Livengood thinks that he is tailoring his normative advice to folk in schools which are similar to the large state school he is familiar with.

"We have an obligation to help our students to pursue their projects and realise their aims." - Livengood #RFMP2019

First questioner describes @Prof_Livengood as an anarchist with a tie - he's right, it's a great tie. #RFMP2019

@Prof_Livengood Now some discussion and disagreement as to the extent to which philosophers actually reason in a deductively valid fashion.

@Prof_Livengood Questioner raises worry: we shouldn't take people's preferences re life projects as given, as philosophers we should hope to shape people's preferences towards the good through teaching. Livengood responds by saying he doesn't feel confident enough to make choices for other folk.

@Prof_Livengood Long discussion now about what we are trying to prepare graduate students for, given the reality of the job market where many graduate students will not be getting academic employment.

@Prof_Livengood Ok folks, we're going to do a roundtable discussion next which I am moderating, so I probably can't liveblog it! There'll be a retrospective blog post over at sootyempiric.blogspot.com to reflect on what I have learned from #RFMP2019 - so bye for now, more later!