/lit/ - Literature

>>15441642
Could you summarize his argument for me? It's been years since I read Hume and one of the side effects of grad school is that I feel like everything I've ever read that's not in my field has been pushed out of my brain.

You should read kant, plato and other brains he was interested in before analyzing himself

>>15441777
Rationalists claim that some ideas come from pure reasoning, and are true regardless of experience. But if we follow the chain of reasoning, we find that these ideas are actually grounded in experience, and to claim that nature is adhering to our ideas is a circular argument. Example: every time a billiard ball hits another, the second moves, so a rationalist might assert such a thing as transfer of momentum, and claim that it is a universal law, but of course this law is based entirely on perception. Nobody is born with any innate ideas about transfer of momentum.

Another way of looking at the same thing is that all our ideas are formulated on the past, and yet they appear to be statements about the future. So what rationalists think are conclusive truths are simply statements of probability - the more A causes the B, the more they are inclined to believe there is cause effect between A and B.

>>15441828
I've read every major philosopher mentioned or alluded to in the Enquiry, apart from malebranche and some minor ones. What exactly am I missing?

>>15441828
Kant came after Hume, dumbass.

>>15441642
he ended the philosophical lines of inquiry into those problems. after hume you continue with physics and probability theory.

>>15441878
So he discovered, like Plato had discovered 2000 years before him, that just because things happen with regularity, it doesn't mean they will happen forever? What a genius.

>>15441878
>Nobody is born with any innate ideas
Fuck, there go instincts to the trash. Biology BTFO.

His argument against miracles is piss poor rethoric, not logic.

His argument against miracles is piss poor rhetoric, not logic.

>>15441907
Interesting that Hume basically described modern Bayesian statistics and science in the Enquiry, suggesting that ALL prior information be considered, and that..

>And, perhaps, our progress in natural philosophy is chiefly retarded by the want of proper experiments and phenomena, which are often discovered by chance, and cannot always be found, when requisite, even by the most diligent and prudent enquiry. As moral philosophy seems hitherto to have received less improvement than either geometry or physics, we may conclude, that, if there be any difference in this respect among these sciences, the difficulties, which obstruct the progress of the former, require superior care and capacity to be surmounted.

>>15441927
Nice reading comprehension.

>>15441933
1. He talks about instincts in the book, which I'm not going to summarise for you.
2. This is a book on epistemology, it has nothing to do with what you're saying. At least learn the basics before spouting shit.

>his skepticism is irrefutable
Do yourself a favour and read Kant

>>15441642
>By the way I have already read Descartes, Spinoza, and Leibniz, so I'm not sure where to turn for a response.
You should turn to the guy who actually responded to Hume and historically won the debate, Immanuel Kant.

>>15441927
that's just a summary brainlet

>>15441959
>Hume basically described modern Bayesian statistics
You aren't wrong. And that's why he's worthless. Bayesian statistics have bad reputation for a reason.

It's not that the concept is wrong in theory. It's that people trying to apply the concept are almost invariably wrong, because it's a problem of information.

So ironically, Hume's method only works in the intangible world of ideas.

One such is example is the Bayesian coronavirus models being all wrong due to the a priori assumptions being worthless. Meanwhile the models without assumptions worked much better.

>>15441878
That's a fine reasoning but how do you explain our ability to formulate general statements? Those statements might be misleading, incomplete or even completely false -yet we do not proceed merely by accumulation of unrelated observations. We bind observations together using principles whose origins is unclear, and over time we get good enough at it that we effectively manipulate real-world events to an extent.

Does Hume have any explanation for that? Even if we concede him our inability to grasp universal laws and the necessity of empirical observations at all step of the cognitive ladder, you should have to account for how that ladder is built in the first place.

This is precisely what Kant attempts to do in Critique of Pure Reason.

>>15441988
That chart is retarded and so are you.

>>15441959
Well, your summary is trash. It should be more specific.

>>15441642
Hume is absolutely right, as far as he goes.

>>15442008
Funny because it's a pro-Bayesian chart, not one meant to criticize it.

Anyway, if we applied Bayes to our history of success and failure with Bayesian models, one should discard Bayes. The irony.

>>15441927
Someone correct me if I'm wrong, as I haven't read Enquiry in a long time, but I believe the thrust of his argument was that the foundation of empirical knowledge is the belief that because certain effects have been always linked with certain causes in the past, then they will continue to be so linked in the future, or that because certain propositions were true in the past, they will continue be true in the future. But the only possible justification for this belief, Hume points out, is that it has been true in the past. Therefore the foundation of empirical knowledge is itself empirical, which is a circular argument like >>15441878 was describing. If Plato had an equivalent argument or observation feel free to enlighten us but I don't think he did.

>>15441988
All models have underlying assumptions, bayesian statistics only make those explicit, and also provide a formula for modifying assumption in light of new observations.

It's also not altogether a bad framework, it provides powerful methods but as with everything in statistics there are condition and limits on the proper use of the method, and even then there's no guarantee it will work well on real data.

Otherwise you're right, bayesian statistics is overused as a meme on the interwebz. It's a nice method among others, nothing more.

Historically Bayes and Hume are contemporary, so it's not surprising similar ideas would pop up in their writings.

>>15442027
There are pure mathematical models that try to predict the curve only based on the shape that is has taken so far. The Bayesian models all invariably include data from previous pandemics or worse.

One such example is how the Italian Bayesian model was worthless because it assumed that by certain date, there should be 0 cases already because the curve in a pandemic "is always symmetric". Even after the symmetry had been broken, it kept trying to restore it by making the cases suddenly drop to zero. The result was insanely wrong predictions every single day.

The pure mathematical model that looked at the shape of curve, and predicted smooth changes instead, was a lot more accurate because it didn't have the a priori assumptions.

>>15441885
Hume read Malebranche very closely, he's actually a major influence. worth not skipping.

>>15442023
>pro-Bayesian
That's like being "pro-arithmetic". Bayes Theorem is a mathematical truth.

>>15442060
Using bad assumptions often leads to bad outcomes. Nothing to do with Bayes.

>>15442060
>There are pure mathematical models that try to predict the curve only based on the shape that is has taken so far. The Bayesian models all invariably include data from previous pandemics or worse.

Not necessarily a bad thing, especially when there are known underlying dynamics (like viral contagion in this case).

>One such example is how the Italian Bayesian model was worthless because it assumed that by certain date, there should be 0 cases already because the curve in a pandemic "is always symmetric".

This is a problem of retarded assumptions, which boils down to a problem of retarded bayesian users, not a problem of the retardedness of bayesian statistics in themselves.
Not that choosing reasonable assumptions is easy. I'm the first to say bayesian stuff is not the scientific magic wand internet bayesian make it out to be. Still bayesian method have been used with success in various fields.
Nothing here contradict what I said above.

>The pure mathematical model that looked at the shape of curve, and predicted smooth changes instead
You can't predict smooth change without some smoothness assumptions in the first place. Even saying "I want a curve that's has a second-order derivative" is an assumption. A bayesian prior is only a very explicit case of assumption.

Even ensemble methods carry with them the underlying assumptions of the algorithm they use, only those assumptions are hard to write out explicitly (random forest for instance still make assumption about the distribution being captured, but those assumptions are encoded in the division criteria used and the hyperparameters).

>>15442094
True, but there's a trend on the internet (and elsewhere) to claim an opposition between "bayesian" and "frequentists". This distinction is not entirely artificial and has philosophical roots, and it often captures the discussion whenever bayesian statistics comes up.

Every professional statistician I've seen or read talking about bayesian statistics (and that include big shots like Van der Waart) are keen to point they're not philosophically bayesian, however, and that they use bayesian statistics as merely a tool among others.

In fact while the internet-bayesian mention bayes' formula a lot the distinction is almost entirely unrelated to the mathematics themselves.

>>15441988
do you actively try to be retarded? if so great job, this is impressive.

>>15442200
>In fact while the internet-bayesian mention bayes' formula a lot the distinction is almost entirely unrelated to the mathematics themselves.
I think the main difference manifests itself in the attitude toward conventional inferential statistics -- i.e., everything one is taught in Stats 101: the T-test, ANOVA, etc. Relying on such methods is fundamentally irrational in cases where relevant prior knowledge exists.

>>15441642
>Did he solve the empiricism vs rationalism debate? I just cannot see a way out for the rationalist.
He destroyed empiricism too, Europe was basically idealist for a couple hundred years after Hume

>>15441991
>This is precisely what Kant attempts to do in Critique of Pure Reason.
Does Kant explicitly discuss the cognitive and psychological faculties that allow us to reason about the perceptions? And where?

>>15442200
>In fact while the internet-bayesian mention bayes' formula a lot the distinction is almost entirely unrelated to the mathematics themselves.
You're confounding two things.

1. Bayesian statistics as a tool used by real statisticians to account for information known by a rational agent but that is not part of the data collected.

2. Bayesian statistics as a paradigm used by internet dorks and capital R Rationalists of the Less Wrong crowds, who employ it to make all sorts of arguments completely outside of the domain of statistics.

>>15441966
>>15441968
Interested. What was Kant's response? Please don't tell me the thing-in-itself or nuomena or whatever, I want a strong argument.

Kant does well in defending Empirical Realism from Humes Skepticism. You should get into the Critique next and see if you buy Kants argument for yourself, nobody on this site will explain it sufficiently to you

>>15442895
Fair enough. Should I go straight into the critique of pure reason or do I need to read anything else first?

>>15443030
If you have Hume down, you can probably just jump into it. I'd recommend reading his Lectures on Metaphysics pre-Critique first to get a groundwork, you can also pick up the Prolegomena which is a shortened summary of the Critique but I don't think it's a necessary pre-req or anything. Use Lecture notes online and some secondary works to get through the Analytic, which is where the main thrust against Hume is located.