Just Above Sunset
Volume 5, Number 10
March 11, 2007

Epistemology Notes

 The world as seen from Just Above Sunset -

"Notes on how things seem from out here in Hollywood..."

Colors We Don't See

The subscription ran out.  The Economist does not appear in the Hollywood mailbox any longer.  The neighbors still get their Variety and Hollywood Reporter, or Rolling Stone - normal here at the edge of the Sunset Strip - but none of them now smirk at the odd guy on the top floor when he unpacks his mailbox in the lobby. The regular newsletter from the ACLU is, of course, acceptable.  Larry Flynt and minor movie stars are involved in that.  But a weekly center-right British magazine - covering international politics, business and world trade, with a few items on culture and quite a few on technology, with its odd want-ads (this or that minor nation looking for a new Director of Economic Policy or some university in Hungary looking for a Dean of Information Studies) - is just too peculiar. Ah well, it was free, sort of - those random airline frequent flier miles, not enough for a trip anywhere, were going to expire and it was a way to make something of them.

But it was fun - the writing was clear, and sometime witty.  In the nineteenth century Walter Bagehot was the editor, and he was droll - "Poverty is an anomaly to rich people; it is very difficult to make out why people who want dinner do not ring the bell."  Or this - "Nothing is more unpleasant than a virtuous person with a mean mind."  It's that British dry quip thing.  The current editors try to carry it on, as best they can.  Hollywood, however, hasn't been like that since David Niven and George Sanders hung around, working at the neighborhood studios. (George Sanders suicide note was famously dry - "I feel I have lived long enough. I am leaving you with your worries in this sweet cesspool - good luck.")

But now that The Economist no long raises eyebrows here, one sees on the web items that are nicely informative, and dryly ironic.  In the January 18 issue there was an item on how "psychologists are learning more about how colour builds language and language builds colour" (yes, with British spelling and all).

Now that's an odd idea, and it opens with this -

    Languages divide the spectrum up in different ways. Welsh speakers use "gwyrdd" (pronounced "goo-irrrth") as a general word for green. Yet "grass" literally translates as "blue straw". That is because the Welsh word for blue ("glas") can accommodate all shades of green. English-speaking anthropologists affectionately squish "green" and "blue" together to call Welsh an example of a "grue" language. A few of them think grue languages are spoken by societies that live up mountains or near the equator because ultraviolet radiation, which is stronger in such places, causes a progressive yellowing of the lens. This, the theory goes, makes the eye less sensitive to short wavelengths (those that correspond to the green and blue parts of the spectrum). Unfortunately, though the Welsh do live in a hilly country, it is hardly mountainous enough - let alone sunny enough - to qualify.

Since the Welsh actor Richard Burton is no longer around - it's been more than forty years since he was working with Fox on Cleopatra (and working on Liz Taylor) over Pico and Motor - we cannot ask him about this Welsh business with the colors.  But it doesn't matter - this ultraviolet theory is just one idea, among many it seems, in the debate about "the psychology of color."

You didn't know there was such a debate? There is, and it's the usual in such circles - on one side congenital, fundamentally genetic, explanations for something, and on the other, explanations that rely on "environmental determinism." The psychologists in the first group think people are born with innate, ingrained ideas about how hues are grouped. The idea here is that the brain seems to be preconditioned to pick out six basic colors no matter what language one is taught to think in.  The other side says no - the spectrum can "be chopped into categories anywhere along its length" and it is the language any individual learns from his parents that determines where that chopping takes place.  In the second view, which color is which is a linguistic construction. There's no such thing as green, per se.  There's just what we say is green.

That leads to what The Economist reports, that two papers published in the Proceedings of the National Academy of Sciences suggest there may be a middle ground.

The post recent of these, from just last week, is from Terry Regier of the University of Chicago.  He and his colleagues worked on the question of "preconditioned" language categories.  They developed a grid displaying all possible hues rolled into a globe, with black at the North Pole and white at the south. It must have been impressive - colors stick out from the sphere according to how sensitive the visual system is to them.  But you really have to think about this -

    Bright yellow, for instance, is easily noticed against a background of other colours, so the yellow part of the sphere bulges. Overall, the knobbly globe has exaggerated, smooth mountains with valleys in between.

    If humans really are hardwired to home in on six focal colours, then all languages should assign words around those six. Dr Regier, however, tests a subtler concept. He thinks that useful languages should allot words in order to minimise the perceptual difference between colours of the same category, and maximise it between colours in different categories. Unlike national boundaries, linguistic boundaries should form only in the valleys of his colour globe, never over the hills.

    Dr Regier therefore programmed his computer to find the best valley borders according to whether he told it to create three, four, five or six "countries" on the globe. Then, to judge whether people build languages around what their brains are best attuned to, he compared these theoretically best divisions with real-world dividing lines.

Fine - and add to that, since 1976, Paul Kay of the UC Berkeley - one of the paper's co-author, has compiled a database of information about how over one hundred different languages assign color adjectives to over three hundred different hues. When this database was compared to the fancy globe thing, "nature: appeared far important than any language in how we name colors.

But it wasn't that simple -

    The model closely fits some languages and points correctly to some details. For instance, three-colour language systems, which lump red and yellow together, generally exclude whitish yellow from that category - as does the model. But the results also explain where nurture gets its wiggle room. Real lexical boundaries tend to vary where Dr Regier's algorithm produced several options that were almost as good as each other.

This is amazing, or these people have too much time on their hands.

But there was a second published last week - by Regier, Kay and now Vicky Drivonikou, of the University of Surrey (UK) - and this one works with the idea that brains have two hemispheres, but most language-processing is done in only one of them.  So that might lead somewhere.

Try this -

    For reasons that lie deep in the evolutionary history of the vertebrates, the right hemisphere deals with sensory input from the left-hand side of the body, and vice versa. It is the left hemisphere, though, that deals with language - at least, it does in right-handed people. If language does affect colour-perception, then it is more likely to affect perceptions from the right visual field than the left.

    In the first phase of the study, the researchers showed that their volunteers (Surrey university students) were slower to notice a target shade of blue when it flashed up against a blue background of a different hue than when it flashed up against a green background. Either nature or nurture (in a non-grue language) might explain that observation. But the researchers went further by presenting the test separately to the left and right visual fields of their volunteers. They found that they got the effect whichever visual field they presented the test to, but it was stronger when the presentation was to the right one (ie, to the left-hand, language-processing side of the brain), than to the left one.

So the non-linguistic side of the brain distinguishes between blue and green just fine, thank you. Language has little to do with any of it, except for the speed at which one makes the distinction. So language really does have something to do with it.  Bah.

It's an old issue. See Jules David Law - The Rhetoric of Empiricism: Language and Perception: From Locke to I.A. Richards - does language create our reality? Can we think about things for which we don't have the words? If we have lots of words for things, can we think better and with more subtlety?  If we have few words - awful language skills - is our ability to think things through severely crippled? (Insert your own George Bush comment.) Do people who have different words - who have radically different languages, with odd grammar and verb tenses we just don't get - actually see differently and think differently?

You could also, if you're a glutton for punishment, read On the Interdependence of Language and Perception (Theoretical Issues In Natural Language Processing, Proceedings of the 1978 workshop on Theoretical Issues in Natural Language Processing) - and the abstract goes like this -

    It is argued that without a connection to the real world via perception, a language system cannot know what it is talking about. Similarly, a perceptual system must have ways of expressing its outputs via a language (spoken, written, gestural or other). The relationship between perception and language is explored, with special attention to the implications of results in language research for our models of vision systems, and vice versa. It is suggested that early language learning is an especially fertile area for this exploration. Within this area, we argue that perceptual data is conceptualized prior to language acquisition according to largely innate strategies, that this conceptualization is in terms of an internal, non-ambiguous "language," that language production from its beginnings to adulthood is a projection of the internal language which selects and highlights the most important portions of internal concepts, and that schemata produced in the sensory/motor world are evolved into schemata to describe abstract worlds. Examples are provided which stress the important of "gestalt" (figure-ground) relationships and projection (3-D to 2 - 1/2 or 2-D, conceptual to linguistic, and linguistic to conceptual); finally mechanisms for an integrated vision-language system are proposed, and some preliminary results are described.

That's Noam Chomsky territory, generative grammar - language of a very basic sort is innate and all that.

On the other hand The Economist item may be a spoof - some sort of dry British satire. The neighbors here may be right. But you can look up the research papers the item cites.

So how do we know what we know - and is there stuff we don't know (or can't with the words we have available), and colors we don't see? Perhaps it's bet not to chat about such things with the neighbors.

This item posted - in its final version - January 28, 2007

[Epistemology Notes]

Last updated Saturday, March 10, 2007, 10:30 pm Pacific Time

All text and photos, unless otherwise noted, Copyright © 2003, 2004, 2005, 2006, 2007 - Alan M. Pavlik