Looking Out To Sea

Some random thoughts on intelligence, language and related matters.

Creating intelligent agents must be done organically

Human babies don’t learn about the world through databases, but through the input of their own senses. So to create a human-equivalent agent it makes sense to give it the equivalent set of sensors and effectors as a human being.

This also suggests that any agent modelled on something else — a spider, a cat, a condor — would be as alien in thought from us as the animal it is designed after.

If you can’t understand how your cat thinks, how can you expect to find something in common with a being whose only knowledge of the world is through a text terminal or a single fixed camera?

Human languages are too opaque for serious use

“Normal” — accepted (behaviour) or average (production) or perpendicular (line)? No wonder people spend so much time arguing.

Shibboleths are words which mark you out as a member of a particular group. What’s the word for a word which different groups have in common, but with wildly different definitions?

Creating a more explicit imperative language

In light of Simon Peyton Jones’ remark of Haskell being an excellent imperative language, what exact combination of “programmable semicolons” is needed to recreate something like C?

newtype C a = C (ReaderT Const (StateT Global IO) a)
    deriving (Functor, Monad, MonadIO, MonadState Global, MonadReader Const)

(The above example is basically just the X monad from the window manage Xmonad.) Is there anything else? Disregarding syntax, is the C monad equivalent to a standard imperative language? (In fact it maybe be more like Python than C, given its higher-levelness.)

Language ambiguity redux

Can we create a similar stack of environments and assumptions for conversation, from more primitive/abstract building blocks? (Obviously, short answer is no. But bear with me.)

More useful would be a type checker for internet arguments that spits out the following when required:

Error: Ambiguous context for keyword `normal' at line 17.

What do you think “make an offer subject to survey” would mean?

1. We get the place valued and then, if the lenders say aye, we make an offer.
2. We make an offer and if the sellers accept it the place is valued and the lenders say aye or nay.
3. We make an offer and get the place valued at the same time. Then the lenders say aye or nay and so do the sellers.

I assumed 2. The lenders seem to prefer 1. The actual state of affairs seems to be 3. Which is the only one that couldn’t be inferred from the text. What the hell? 1 and 3 are also the only ones that penalise the buyer if they don’t get accepted by the seller.

I hate house-buying. :-(

I’ve just had a look at http://dictionary.reference.com/search?q=believe for the meaning of the verb ‘to believe’ and I found a number of very interesting definitions — to have confidence or faith in, to hold true, to suppose or assume — but none of the definitions fit this sentence (my emphasis):

Last week we learned that a family of at least three children had come down with measles because their family didn’t believe in vaccinating them …

There are other examples — just think of all the things people object to. “I don’t believe in sex before marriage”, “I don’t believe in abortion”, “I don’t believe in drugs”.

It should be fairly obvious that this type of person does actually believe in the existence of vaccination, pre-marital sex, abortion or whatever. They might even be involved in protests against them. But it’s not that they don’t actually believe in them. It’s that they don’t approve of them. So why do people use ‘believe’ when they mean ‘approve’?

And just as interesting, why does this very common usage not appear in dictionaries? (I also tried Urban Dictionary and Wiktionary, two sources I thought might mention ‘unofficial’ usage. But nothing.)

All this brings up some interesting thoughts when people make “belief” claims. If an otherwise smart person says “I don’t believe in evolution”, what do they mean by that? Many creationists argue against evolution on the grounds of moral consequences — that Hitler was an evolutionist, that Darwinism is a cruel and inhumane philosophy leading to genocide and eugenics, etc. It’s perfectly possible that “I don’t believe in evolution” has a strong element of “I don’t approve of evolution” or “I don’t like the moral consequences of evolution”.

And since few people seem to make a clear distinction between “belief as faith” and “belief as approval”, there are further consequences. Is “belief in God” entirely to do with faith, or is there an element of approval? A case of “I approve of God, because that is how I would like things to be”. An important element of religious faith is the comfort that people derive from it.

I have to admit this is all idle conjecture. But one final thought: whenever I find myself trying to find support for a belief of my own, I have to ask myself why it’s so important. And the only honest answer I can give myself is that, it’s important because this is the way I’d like things to be. If an important part of my world-view is removed, then I have to re-evaluate it all for consistency. That is a lot of work and may reveal things I don’t like — about myself, my friends or life in general.

I’m curious whether these memes have any basis in reality:

• Men have better visual/spatial reasoning than women.
• Women have better language skills than men.
• Men have better abstract reasoning skills than women.
• Women are better at multitasking than men.

I’ve googled lots but come up empty. Well, lots of people convinced of the truth of these statements but no-one with so much as a reference to follow up.

The women/language one is the only one I know a little about, in that the Daily Mail seems extraordinarily keen on repeating the pop-science myths which The Female Brain described — women talk more than men, etc.

Some information on the origins and veracity of these claims would be really interesting.

I’ve got to record what we did this week at BSL because Helen wasn’t there. And we’ve got homework so I thought I’d write it here so it’s easy to access.

Describing people and animals

We followed on from the previous week, doing more description of animals first, then of people after the break. It’s still really hard to picture what is distinctive about people. Same with animals — there’s always something that’s really stereotypical or indicative but it’s hard to come up with. Last week I was trying to describe a wolf but completely forgot about the “howling at the moon” behaviour, so nobody knew what I was talking about.

This week I managed to convey a goat pretty well, but it’s easier because not many animals butt heads and have long beards. That’s really goatish. I then elected to describe Johnny Rotten, which I thought would be pretty straight-forward. The first guess was Sid Vicious! Meh.

Still pretty difficult to describe without using topic words: to say “23” instead of “age 23” or “blonde” instead of “hair blonde”. It’s the same mindset that has people prefixing every sentence by pointing to themselves. In other words, English grammar.

Holiday

We spent the majority of the lesson covering arrangements for holidays — stuff you need to have, stuff you need to do, that kind of thing. Most of it airport-based, I suppose because it has the most scope for distinctive vocabulary. People use trains to get to work, but departure lounges and x-ray scanners are not day-to-day things.

Let me see, there was “waiting” and “queue” and “lounge” and “complain” and “argument”. You can tell we’ve all been to airports before, right? And “late” but also “early”, “sunglasses” and “midge repellent” and the names of a bunch of different countries.

The sign for Hawaii is good, because it’s like a little grass-skirted hula dance. For the record, Scotland is a stylised bagpipe sign, a sort of one-armed chicken flap. This is apparently called metonymy — referring to one thing by describing something that is related to it. English has this when we talk about “the crown” to mean the monarchy. This happens loads in sign language and is one of the things I really enjoy about it. Some of the examples I’ve seen have been very enlightening and evocative (though naturally I can’t think of any right now).

Homework

For next week we’ve to come up with a two-minute story describing “setting off on holiday”. Everything to the point where the plane takes off, except the decision-making process which is boring. So: packing, readying the house (cancelling milk, etc.), getting money/insurance, leaving for the airport. That should be more than enough for two minutes of presentation.

In Guns, Germs and Steel (a very interesting though eventually quite repetitive book), Jared Diamond tells of a native American who invented an alphabet for his own tongue, after seeing how the Europeans gained such power and expression from their written words. The story (as I remember it) is that he got a sample of the alphabet we used and started assigning sounds from his own language. When he reached the limit of the borrowed glyphs, he started to invent new ones.

It seems odd that people think of the alphabet we use in this country as ‘the English alphabet’ when it so poorly suited to the English language. So many sounds we make in English are not catered for — we must use combinations like ‘sh’ or ‘th’ or ‘ch’ to make up for the shortfall Over the centuries we’ve even lost letters that used to do some of these jobs. Even if we consider the stand-ins as single letters rather than pairs, there are so many odd exceptions and disparities between spoken and written English. Would you guess ‘onion’ is actually pronounced more like ‘union’ looks, which itself is actually pronounced more like ‘yoonyin’? (But then, you’ll probably disagree completely because your accent is so different from mine.)

There is an interesting article about the effect of this arbitrary mapping of sound to symbol. The awkwardness of the English language in this regard is contrasted with reading in Germany and Austria, with terrifying results:

Some children in countries with transparent alphabets do have reading problems, but these have to do with fluency and comprehension. Yet even this is relative. Normal readers from Salzburg were compared to normal readers from London on tests of reading accuracy and speed. Seven year-olds from Salzburg read as fast as the 9 year olds from London, making half the number of errors. The Austrian 7 year-olds had one year of reading instruction, the English 9 year-olds, four or five.

There seems to be a long history of attempted spelling reform for the English language, with limited success. The American dictionary compilers have obviously had the most success, though all they’ve really done is entrench differences in pronunciation. There is a big difference between ‘mum’ and ‘mom’, after all.

I started along this track after reading complaints from BSL users about using the English alphabet. BSL reduced to written words removes all the subtlety and all the power from the language. But then, it does that to most spoken English as well. I wonder if the American fellow mentioned above, who invented his own alphabet to suit his own way of speaking, felt the same?