March 18th, 2009

Captive Dreams

(no subject)

Do Computers Think, or Do Thinkers Compute?

Here's a thought from James Chastek that is relevant to discussions of whether computers can ever think.  It all boils down -- as so many things do -- to Latin grammar:

In classical Latin, only principle agents are said to perform actions, and all instruments were put in the ablative of instrument. For example, if you wrote a sentence like “This airplane can fly you from Paris to New York in four hours”, a grammarian would tell you that you had to correct it to read something like “A pilot can fly you from Paris to New York in four hours with this plane”. If an advertisement read “this corkscrew removes corks better than the leading brand!” all educated people would sigh. A person using this corkscrew (or by this corkscrew) could remove bottle tops,  etc. The rule still applies in German.

I’m reminded of this rule in thinking about claims in philosophy of mind, many of which are simply bad grammar: “A scanning MRI can tell your intentions”; “the brain thinks”; “can a computer think?”  etc.

Organs and computers are instruments. That’s just the way it is. If someone can’t tell the difference between an instrument and and a non-instrument, he has problems that he has to sort out well before he starts trying to answer questions about the nature of thought.

The philosopher John Searle once illustrated the impossibility of AI by postulating the Chinese Room plato.stanford.edu/entries/chinese-room/ .  (Turing's actual point was similar: there is ultimately no way you can =prove= that the person you are talking to is a person with understanding or simply a very clever construct.  He never proposed that the construct had understanding of the sentences it produced.  But today, one can find enthusiasts even of "intelligent parrots"!!)  Critics of Searle have said that even if the man inside the room does not understand Chinese, but only follows English instructions for manipulating strings of Chinese characters, that "somehow" the "room-as-a-whole" understands Chinese.  Thus, the absurdities to which AI proponents will go.  The "room-as-a-whole" somehow "understands"?  Far easier it is to suppose that the seat of understanding here the person who wrote the English instructions which the man-in-the-room follows and prepared the cards with the Chinese characters. 

Of course, Searle also said: 

`[O]nce we escape the clutches of two thousand years of dualism, we can see that consciousness is a biological phenomenon like any other and ultimately our understanding out it is most likely to come through biological investigation''

But dualism is a Cartesian thing, and hardly 400 years old.  The 2000-year old thing is hylomorphism, the inextricable unity of matter and form.  Consciousness (which is not the same thing as understanding) is likely what we nowadays call an "emergent property" but which in Aristotelian terms did not require a peculiar name.  But our understanding of it is unlikely to emerge from biological investigation so long as biology seeks to emulate old-style reductionist physics.  Just as the gaseous properties of chlorine emerge from its form (the simple number and arrangement of its parts) and not from the powers of its constituent protons, electrons, etc., the emergence of consiousness is more likely a property of the substantial form of the human being and not from the powers of constituent neurons or molecules or atoms. 

+ + +

Stay Frosty, People

Here is an analysis of the beryllium-10 in the Dye-3 ice core from Greenland.  Be10 being rare, it is unlikely to reflect any contamination. 

  • Be-10 is produced by cosmic ray collisions with atoms of oxygen and nitrogen. . 
  • Galactic cosmic rays increase or decrease because the solar wind gets weaker or stronger. 
  • The solar wind gets stronger because the Sun’s magnetic field gets stronger. 
  • So the Be-10 concentrations are related to cosmic ray intensity, thence to solar wind, thence tor solar strength. 

 


The known cold periods correspond quite well to peaks in the Be-10 data, which would be periods of weak solar magnetic field strength.  It even picks out the cold snap around 1900, when Niagara Falls froze solid (1911).  Since that time, Be-10 has been steadily decreasing (sun getting stronger, earth getting warmer) save for one episode from about 1940-1970 when we worried for a moment about global cooling.  It's that blip on the right. 

Why this is worrisome is the following graph of direct measurement of the planetary-solar magnetic index: 


which shows a rather sudden drop in field strength in lat 2005.  This means the field strength is dropping, hence more cosmic rays getting through (and Be-10 being produced) and, gauging by past history, a colder time ahead. 

According to NASA, we are seeing:

• The lowest sustained solar radio flux since the F10.7 proxy was created in 1947;

• Solar wind is the lowest observed since the beginning of the space age;

• The solar wind magnetic field 36 percent weaker than during the minimum of Solar Cycle 23;

• Effectively no sunspots;

• Cosmic rays at near record-high levels.

If this keeps up, we could return to conditions of 200 years ago, when ox-drawn wagon trains could cross the Mississippi on the two-foot thick ice of December.