I’m of the opinion that deductive logic is way over-emphasized in philosophy training, and it seems like it’s even more over-emphasized than usual in your department. I actually think a lot of pathologies in the way mainstream analytic philosophy is practiced are partially attributable to the fact that philosophers learn so much logic and basically no other rigorous analytic tools. Take, for example, the fact that philosophers are usually very careful about whether their conclusions follow from their premises, but the way the premises themselves are justified is often a mess (appeals to intuition, dubious thought experiments, etc). If philosophers knew more about how to infer beliefs from evidence, their arguments would in general be much better.
Even departments that have courses in inductive logic usually treat it as an adjunct. The message philosophy students get is that their primary tool is deductive logic, and they should have a thorough grounding in that discipline, but it’s also good if they pick up some elementary statistics and probability theory. I would recommend the opposite emphasis, but even equal treatment of inductive and deductive logic would be a massive improvement.
I strongly agree with your comment. What concrete steps would you take to fix the problem? Are there specific classes you would add or things you would emphasize in existing classes? Are there specific classes that you would remove or things you would de-emphasize in existing classes?
Some sort of “programming for philosophers” class, teaching basic programming but emphasizing the connections with the material they’ve learned in their logic classes.
I’m tempted to merge symbolic logic and mathematical logic. I’m not sure how they are different from each other. How would you divide them?
I’d put Probability theory before scientific reasoning, since the latter flows from the former (not historically, but this is a philosophy course, not a history one). The result should naturally include the “Bayesian epistemology” part of your formal epistemology course.
Typically, symbolic logic classes focus on reasoning with formal systems and mathematical logic classes focus more on reasoning about formal systems. So in a symbolic logic class you would mainly learn how to do proofs while in a mathematical logic class you would learn about things like Godel’s theorems. Maybe “mathematical logic” is a bit of a misnomer, but it is the traditional title of these classes.
If you agree here, I’m curious why you’re focusing on reforming the logic curriculum? Why not focus on shifting resources from teaching logic to teaching the standard things recommended here (probability theory, heuristics and biases, psychology etc.).
Because I see those things as part of logic. As I see it, logic as typically taught in mathematics and philosophy departments from 1950 on dropped at least half of what logic is supposed to be about. People like Church taught philosophers to think that logic is about having a formal, deductive calculus, not about the norms of reasoning. I think that’s a mistake. So, in reforming the logic curriculum, I think one goal should be to restore something that has been lost: interest in norms of reasoning across the board.
Hmm, is that best accomplished by trying to reappropriate the word ‘logic’? Mathematicians, philosophers etc. seem like they have a pretty firm idea about they mean by ‘logic’ and going against is probably hard. Trying to get a Heuristics and Biases or statistics course into the logic curriculum seems like it would get a lot of pushback. Can the word ‘logic’ itself be that valuable? Why not pick a new word?
I don’t want to have a dispute about words. What I mean when I talk about the logic curriculum in my department, I have in mind the broader term. The entry-level course in logic does have some probability/statistics content, already. There isn’t a sub-program in logic, like a minor or anything, that has a structural component for anyone to fight about. I would like to see more courses dedicated to probability and induction from a philosophical perspective. But if I get that, then I’m not going to fight about the word “logic.” I’d be happy to take a more generic label, like CMU’s logic, computation, and methodology.
I think part of why I think I’m confused is that none of the courses you proposed are focused on psychology (heuristics and biases being the standard recommendation). Any reason for that?
That’s a good point. Looks like an oversight on my part. I was probably overly focused on the formal side that aims to describe normatively correct reasoning. (Even doing that, I missed some things, e.g. decision theory.) I hope to write up a more detailed, concrete, and positive proposal in the next couple of days. I will include at least one—and probably two—courses that look at failures of good reasoning in that recommendation.
Another thing that comes to mind, is that if you’re advising the curriculum committee and not directly in charge, you may want to strategize about how best to convince them to take a more lesswrongy attitude. Things that spring to mind:
getting multiple people to say similar things
making the same argument repeatedly and in private with members of the committee
getting a speaker (luke?) to come in and make the case
I’m of the opinion that deductive logic is way over-emphasized in philosophy training, and it seems like it’s even more over-emphasized than usual in your department. I actually think a lot of pathologies in the way mainstream analytic philosophy is practiced are partially attributable to the fact that philosophers learn so much logic and basically no other rigorous analytic tools. Take, for example, the fact that philosophers are usually very careful about whether their conclusions follow from their premises, but the way the premises themselves are justified is often a mess (appeals to intuition, dubious thought experiments, etc). If philosophers knew more about how to infer beliefs from evidence, their arguments would in general be much better.
Even departments that have courses in inductive logic usually treat it as an adjunct. The message philosophy students get is that their primary tool is deductive logic, and they should have a thorough grounding in that discipline, but it’s also good if they pick up some elementary statistics and probability theory. I would recommend the opposite emphasis, but even equal treatment of inductive and deductive logic would be a massive improvement.
I strongly agree with your comment. What concrete steps would you take to fix the problem? Are there specific classes you would add or things you would emphasize in existing classes? Are there specific classes that you would remove or things you would de-emphasize in existing classes?
These would be the formal classes in my ideal philosophy curriculum:
Symbolic logic (sentential and predicate logic, some model theory)
Set theory and category theory
Mathematical logic (along the lines of your 454)
Scientific reasoning (elementary statistics, causal inference)
Probability theory and the philosophy of probability
Rational decision-making (decision theory, heuristics and biases)
Formal epistemology (Bayesian epistemology, confirmation theory, computational learning theory)
Some sort of “programming for philosophers” class, teaching basic programming but emphasizing the connections with the material they’ve learned in their logic classes.
I’m tempted to merge symbolic logic and mathematical logic. I’m not sure how they are different from each other. How would you divide them?
I’d put Probability theory before scientific reasoning, since the latter flows from the former (not historically, but this is a philosophy course, not a history one). The result should naturally include the “Bayesian epistemology” part of your formal epistemology course.
Typically, symbolic logic classes focus on reasoning with formal systems and mathematical logic classes focus more on reasoning about formal systems. So in a symbolic logic class you would mainly learn how to do proofs while in a mathematical logic class you would learn about things like Godel’s theorems. Maybe “mathematical logic” is a bit of a misnomer, but it is the traditional title of these classes.
If you agree here, I’m curious why you’re focusing on reforming the logic curriculum? Why not focus on shifting resources from teaching logic to teaching the standard things recommended here (probability theory, heuristics and biases, psychology etc.).
Because I see those things as part of logic. As I see it, logic as typically taught in mathematics and philosophy departments from 1950 on dropped at least half of what logic is supposed to be about. People like Church taught philosophers to think that logic is about having a formal, deductive calculus, not about the norms of reasoning. I think that’s a mistake. So, in reforming the logic curriculum, I think one goal should be to restore something that has been lost: interest in norms of reasoning across the board.
Hmm, is that best accomplished by trying to reappropriate the word ‘logic’? Mathematicians, philosophers etc. seem like they have a pretty firm idea about they mean by ‘logic’ and going against is probably hard. Trying to get a Heuristics and Biases or statistics course into the logic curriculum seems like it would get a lot of pushback. Can the word ‘logic’ itself be that valuable? Why not pick a new word?
I don’t want to have a dispute about words. What I mean when I talk about the logic curriculum in my department, I have in mind the broader term. The entry-level course in logic does have some probability/statistics content, already. There isn’t a sub-program in logic, like a minor or anything, that has a structural component for anyone to fight about. I would like to see more courses dedicated to probability and induction from a philosophical perspective. But if I get that, then I’m not going to fight about the word “logic.” I’d be happy to take a more generic label, like CMU’s logic, computation, and methodology.
Ah, okay, that makes sense then.
I think part of why I think I’m confused is that none of the courses you proposed are focused on psychology (heuristics and biases being the standard recommendation). Any reason for that?
That’s a good point. Looks like an oversight on my part. I was probably overly focused on the formal side that aims to describe normatively correct reasoning. (Even doing that, I missed some things, e.g. decision theory.) I hope to write up a more detailed, concrete, and positive proposal in the next couple of days. I will include at least one—and probably two—courses that look at failures of good reasoning in that recommendation.
I look forward to it :)
Another thing that comes to mind, is that if you’re advising the curriculum committee and not directly in charge, you may want to strategize about how best to convince them to take a more lesswrongy attitude. Things that spring to mind:
getting multiple people to say similar things
making the same argument repeatedly and in private with members of the committee
getting a speaker (luke?) to come in and make the case
finding articles that make the same points you do
Strongly agree.