I agree that modern science provides valuable insights into philosophical problems. I also agree that Bayesian probability theory and machine learning are powerful models for approaching problems in epistemology. This is why I’m in grad school in machine learning, and not for philosophy. Furthermore, I’m not a big fan of ancient philosophers (especially ones who think categories are absolute), and I’d like to see the computational theory of mind excised from popular thought, in favor of something closer to embodied cognition. I actually really like the idea of incorporating modern theories and empirical discoveries into a philosophical curriculum.
Despite this, I have a strong negative reaction to your post, because it suggests there is One True Way to do philosophy and that everyone who does not follow the Ways of Bayes is doing it wrong. The last thing I want us teaching students is any kind of absolutism. It can only damage students to tell them that our current models are the true models, and all past thinkers were necessarily wrong. It would also damage students to restrict them to one philosophical viewpoint; as much as I like Bayesian reasoning and empiricism, I think it would hurt students to teach them that these methods are the One True Way, because it would prevent them from exploring alternative viewpoints.
I think that students of philosophy should be taught as many theories as possible, both ancient and modern. By coming to understand the diverse range of models that we’ve applied over the course of human history, students can learn some humility. Just as all of these past models were superceded, our current theories will inevitably be replaced. Just as we can spot the glaring errors in past philosophical models, the people of the future will spot the “obvious” follies in our own ideas.
Also, the more models that students learn, the more “degrees of freedom” they will realize exist. They will come to understand along which dimensions worldviews can vary; they can then explore other options for these dimensions, or discover new dimensions that no one has tried varying yet. I strongly believe that learning more worldviews is a powerful method of keeping one’s mind flexible enough to come up with genuinely new ideas.
Lastly, as much as I love mathematical models and rigorous empiricism, I oppose the trend of applying them haphazardly to the social sciences. If we’re studying e.g. anthropology, I think it’s a mistake to favor statistical data over first-hand accounts or subjective analyses. Not because there’s anything inherently wrong with empirical and statistical methods, but because the models we use are too simple. There are so many features, and it’s hard to account for all of them, both because we don’t know which features to choose, and because inference is computationally intractable in such an enormous model. Fortunately, the typical human brain comes prepackaged with empathy and a theory of mind, a powerful module for modeling the behaviors/preferences/internal experiences of other humans. Certainly, this module is subject to biases and might make systematic errors when reasoning. But when choosing between two imperfect models, I tend to think our built-in circuitry is better suited for the social sciences than tools of machine learning. I assume that our built-in intuitive machinery is useful for some branches of philosophy as well.
Teach the best case that there is for each of several popular opinions. Give the students assignments about the interactions of these different opinions, and let/require the students the students to debate which ones are best, but don’t give a one-sided approach.
Best for what? The idea that you can solve philosphical problems with science is lacking in examples of philosophical problems that have been solved with science. There is just the hope that methods that have worked in one field will work with another.
It grieves me to note that almost all the arguments in your post could be applied, mutatis mutandis, to why we should teach kids intelligent design as well as evolution.
In certain contexts, yeah. I think most kids would be able to smell the bullshit if both theories were laid out side-by-side, with historical context available.
I agree that modern science provides valuable insights into philosophical problems. I also agree that Bayesian probability theory and machine learning are powerful models for approaching problems in epistemology. This is why I’m in grad school in machine learning, and not for philosophy. Furthermore, I’m not a big fan of ancient philosophers (especially ones who think categories are absolute), and I’d like to see the computational theory of mind excised from popular thought, in favor of something closer to embodied cognition. I actually really like the idea of incorporating modern theories and empirical discoveries into a philosophical curriculum.
Despite this, I have a strong negative reaction to your post, because it suggests there is One True Way to do philosophy and that everyone who does not follow the Ways of Bayes is doing it wrong. The last thing I want us teaching students is any kind of absolutism. It can only damage students to tell them that our current models are the true models, and all past thinkers were necessarily wrong. It would also damage students to restrict them to one philosophical viewpoint; as much as I like Bayesian reasoning and empiricism, I think it would hurt students to teach them that these methods are the One True Way, because it would prevent them from exploring alternative viewpoints.
I think that students of philosophy should be taught as many theories as possible, both ancient and modern. By coming to understand the diverse range of models that we’ve applied over the course of human history, students can learn some humility. Just as all of these past models were superceded, our current theories will inevitably be replaced. Just as we can spot the glaring errors in past philosophical models, the people of the future will spot the “obvious” follies in our own ideas.
Also, the more models that students learn, the more “degrees of freedom” they will realize exist. They will come to understand along which dimensions worldviews can vary; they can then explore other options for these dimensions, or discover new dimensions that no one has tried varying yet. I strongly believe that learning more worldviews is a powerful method of keeping one’s mind flexible enough to come up with genuinely new ideas.
Lastly, as much as I love mathematical models and rigorous empiricism, I oppose the trend of applying them haphazardly to the social sciences. If we’re studying e.g. anthropology, I think it’s a mistake to favor statistical data over first-hand accounts or subjective analyses. Not because there’s anything inherently wrong with empirical and statistical methods, but because the models we use are too simple. There are so many features, and it’s hard to account for all of them, both because we don’t know which features to choose, and because inference is computationally intractable in such an enormous model. Fortunately, the typical human brain comes prepackaged with empathy and a theory of mind, a powerful module for modeling the behaviors/preferences/internal experiences of other humans. Certainly, this module is subject to biases and might make systematic errors when reasoning. But when choosing between two imperfect models, I tend to think our built-in circuitry is better suited for the social sciences than tools of machine learning. I assume that our built-in intuitive machinery is useful for some branches of philosophy as well.
You are not supposed to teach them it’s the One True Way, just that it’s The Best Way Anyone Have Found So Far By A Fair Margin.
This also seems problematic, for the same reasons.
And what if it is? I am not claiming this is so. It is rhetorical. What then?
Teach the best case that there is for each of several popular opinions. Give the students assignments about the interactions of these different opinions, and let/require the students the students to debate which ones are best, but don’t give a one-sided approach.
Best for what? The idea that you can solve philosphical problems with science is lacking in examples of philosophical problems that have been solved with science. There is just the hope that methods that have worked in one field will work with another.
It grieves me to note that almost all the arguments in your post could be applied, mutatis mutandis, to why we should teach kids intelligent design as well as evolution.
In certain contexts, yeah. I think most kids would be able to smell the bullshit if both theories were laid out side-by-side, with historical context available.
Illusion of transparency? I believe some creationists say the same thing.
Sounds like an opportunity for a variation on the old Priests of Baal routine, then.