Another probability question, this time with Ravens

[srw]

This whole thread crosses philosophy, formal logic, everyday English and statistics. The latter three are languages we use to talk about the real world, and we need to be sure which language we're talking, while philosophy tries to elucidate the difference. The "paradox" referred to in the OP rests on a confusion between everyday English and formal logic; the question in the OP rests on a confusion between statistics and everyday English - as @swansonj suggests you can't really use statistics to prove propositions that are expressed in everyday English as universals, because statistics always leaves room for uncertainty. You can use statistics (Bayesian or otherwise) to show that within a margin of error all ravens are black, but never that all ravens are black. Anyone who's studied undergraduate philosophy will remember, as @User14044 hints, David Hume's discussion of whether we can say that all swans are white.

(I did formal logic and philosophy as an undergraduate, including a course on Hume, and came late to statistics in my professional exams. I started learning English much earlier.)

In the meantime, here's some music.

 

[swansonj]

This whole thread crosses philosophy, formal logic, everyday English and statistics. The latter three are languages we use to talk about the real world, and we need to be sure which language we're talking, while philosophy tries to elucidate the difference. The "paradox" referred to in the OP rests on a confusion between everyday English and formal logic; the question in the OP rests on a confusion between statistics and everyday English - as @swansonj suggests you can't really use statistics to prove propositions that are expressed in everyday English as universals, because statistics always leaves room for uncertainty. You can use statistics (Bayesian or otherwise) to show that within a margin of error all ravens are black, but never that all ravens are black. Anyone who's studied undergraduate philosophy will remember, as @User14044 hints, David Hume's discussion of whether we can say that all swans are white.

(I did formal logic and philosophy as an undergraduate, including a course on Hume, and came late to statistics in my professional exams. I started learning English much earlier.)

In the meantime, here's some music.

It's true that embedded within statistics is always a concept of uncertainty. But that wasn't the main point I was trying to make.

Because statistics appears to be a pretty objective process - you take data, you plug them into a formula, the formula yields the answer - there is a widespread impression that it is indeed objective. That is misleading, because there is a vast (and often unrecognised) subjectivity involved in formulating the hypotheses you are testing. Every time someone uses p=0.05, or decides on a one-sided or two-sided test, or indeed decides which of two alternatives they will treat as the null hypothesis, they are making a (small "p") political choice.

This maters enormously when it comes to looking at evidence for health risks, or the effectiveness of health treatments, etc. The person who gets to choose the test to be applied can often pretty much determine the outcome before the first number is ever entered in a spreadsheet.

 

[srw]

It's true that embedded within statistics is always a concept of uncertainty. But that wasn't the main point I was trying to make.

Because statistics appears to be a pretty objective process - you take data, you plug them into a formula, the formula yields the answer - there is a widespread impression that it is indeed objective. That is misleading, because there is a vast (and often unrecognised) subjectivity involved in formulating the hypotheses you are testing. Every time someone uses p=0.05, or decides on a one-sided or two-sided test, or indeed decides which of two alternatives they will treat as the null hypothesis, they are making a (small "p") political choice.

This maters enormously when it comes to looking at evidence for health risks, or the effectiveness of health treatments, etc. The person who gets to choose the test to be applied can often pretty much determine the outcome before the first number is ever entered in a spreadsheet.

Philosophy of statistics! All we need now is a dose of consideration of the infinite and someone asking "but what is a number?", and we've got a complete course in maths and philosophy.

 

[twentysix by twentyfive]

David Hume's discussion of whether we can say that all swans are white.

Australian swans (Cygnus atratus) would disagree that they are white. But some b*gg*rs do claim that black is white (see SC&P)

 

[srw]

Australian swans (Cygnus atratus) would disagree that they are white.

C. atratus were unknown when Hume was writing. Hence the "Black Swan" beloved of Naseem Taleb.

 

[Reynard]

But not all ravens are black.

https://www.redbubble.com/people/ravenprints/works/1877692-dare-to-be-different-rare-white-raven

You forgot to add genetics into the mix Though that does also involve statistics. Though statistics and the quirks of mother nature don't always co-operate

That raven is white, though not an albino. Those are two separate and completely different mutations, both of which do occur naturally. An albino would have pink eyes and has no colour pigment, though to our eyes it still appears white. White (as opposed to white spotting) in animal colouration is usually a "masking" gene as it will take precedence over any other colour genes the animal may have.

In the wild, white animals are far less less common than they should be, simply because they are targets for bullying and predation due to their colour. They stick out like the proverbial sore thumb. That is, of course, excepting those who are white to blend in with specific environments.

If you recorded the individual genomes of a population of ravens, you'd be able to calculate the probability of getting a non-black bird. But since chance dictates which sperm joins with which egg, you still might not hatch a white one unless the parents are either both white or both albino, as generally the genes for a change in an individual's colour (certainly as far as cats are concerned, but would imagine the same is true of birds) are recessive. This means that both parents need to carry at least a single copy of said gene for the trait to be passed to the offspring.

Case in point, a cat breeder friend of mine mated a chocolate queen to a lilac stud. The queen carries the gene for lilac and both carry the gene for colourpoint. She was expecting to get a litter of mainly lilac kittens with perhaps one or two chocolate and colourpoints. She ended up with a litter of six chocolate kittens.

 

[twentysix by twentyfive]

C. atratus were unknown when Hume was writing.

Indeed.

 

[Reynard]

Albinism (which is a *lack* of pigment) is purely genetic, nothing to do with diet. It is also a different gene for the one that is responsible for masking white. The former will have red eyes, the latter will not.

In contrast, the pink in flamingoes is to do with diet - namely the shrimps that they eat. Captive birds have the pigment added to their food, else they'll be white and not pink.

That blackbird - that's a spotting gene, which is independent of the gene for all white. I had one of these chaps in my garden a year or two back, but the spotting was on his chest, not on the wings.

HTH

 

[Reynard]

Hmmmm. The RSPB web page doesn't really differentiate between the three different - and distinct - genes for causing white and vitiligo.

Admittedly my experience with colour genetics is with cats, but many of the genes have similar effects across the species. Though I know diddly squat about birds that doesn't involve bacon lattices, sage and onion...

 

[Reynard]

[QUOTE 4769560, member: 9609"]I know very very little about it, hence all my question marks in my earlier post. But it would appear from what I have been reading tonight that diet plays quiet a big part in how birds create colour, much more so than in mamals, hence white patches can come and go if the diet is poor. However it also sort of reads that full albino is a genetic thing. And as you say partial albino is quite a different kettle of fish to full albino but some sites seem to group it all together. Sometimes the more I read the more I get confused and this is probably one of those subjects.[/QUOTE]

We both learnt something. Win win

 

[ufkacbln]

Hmmmm. The RSPB web page doesn't really differentiate between the three different - and distinct - genes for causing white and vitiligo.

Admittedly my experience with colour genetics is with cats, but many of the genes have similar effects across the species. Though I know diddly squat about birds that doesn't involve bacon lattices, sage and onion...

Some birds are reliant on dietary pigment

Flamingos are a case in point

 

[Reynard]

In contrast, the pink in flamingoes is to do with diet - namely the shrimps that they eat. Captive birds have the pigment added to their food, else they'll be white and not pink.

Point already covered. Learnt that one from watching Animal Park