REAL PATTERNS

The notion of a pattern

A pattern is real if there is an algorithm that describes the sequence in fewer bits than it would take to describe the pattern element by element.

a. Whether a pattern is there is an objective fact, whether we can discern that fact or not.

b. Not all patterns are visible to us; in fact, we can assume that very few patterns are actually made available to us through our sense organs.

c. Nor need anyone be able to recognize the pattern at first glance.

d. Whether or not we see a pattern is often dependent upon the concepts we have —they serve to organize the data for us. Chess game.

Using Patterns.

People use patterns in different ways, for different purposes. Three different attitudes towards patterns.

We want to be able to make bit map reduplication of the pattern.

We care about the noise but not where it occurs.

We want to understand the “underlying pattern” and simply ignore the noise.

The Jones and Brown example.

Jones. Simple pattern with high noise.

Brown. Complex pattern with low noise.

Both the strategies of Jones and Brown offer improvements over the bit pattern; one is simple but not too reliable; the second is very reliable but work intensive. Which patterns Jones and Brown choose is a pragmatic decision based upon their preferences and needs. If Jones is playing poker on line for fun and amusement, he will be happy beating the odds, using a simple pattern with high noise. If Brown is a professional on-line player who depends upon this game for his income, he will be willing to put in the effort in order to be more consistently correct.

Levels of Patterns: The Game of Life

Three Basic Points from The Game of Life.

1 Out of very simple rules, rules that are not even represented explicitly but merely hard-wired into a mechanism, very complex behavior can be produced.

2 The elements, their properties and the rules that govern the elements change from level to level.

3 The noise in the pattern increases as the complexity grows.

In the Game of Life, we start at the most simple, level and then progress…

a. individual pixels (which turn on and off)

b. entities composed of pixels, or objects such as gliders, eaters (that move and ‘attack’, are ‘eaten’, that ‘smash’ or break up other objects, that persist through time

c. Turing Machine moves (which embody machine states and the transitions between them)

d. E.g. Chess moves — the intentional level of “moving out the queen early”, etc. Here the intentional level depends upon whatever programme the Turing machine is running.

At the top level, these will be intellectual patterns, not visual patterns, in which see behaviors as being certain kinds of actions.

“Note that there has been a distinct ontological shift a we move between levels; whereas at the physical level there is no motion, and only the individuals, cells, are defined by their fixed spatial location, at this design level we have the motion of persisting objects; it is one and the same glider that has moved southeast…and there is one less glider in the world after the eater has eaten it…”

“..in the ontology at this level; their salience as real things is considerable but not guaranteed. To ay that their salience is considerable is to say that one can, with some small risk, ascend to this design level, adopt its ontology, and proceed to predict— sketchily and riskily—the behavior of large configurations or systems of configurations, without bothering to compute the physical level.”

When are patterns real? The Reality of Intentional Patterns

An obvious assumption when seeing a complex organism such as people or even personal computers is to assume that the intentional level exists because somewhere in the organism there exist propositional-like units that cause the complex actions to occur.

Industrial Strength Realism: “For Fodor, the pattern seen through the noise by every day folk psychologists would tell us nothing about reality, unless it, and noise, had the following explanation: when we discern from the perspective of folk psychology is the net effect of two processes: an ulterior hidden process wherein the pattern exists quite pure, overlaid and partically obscured by the various intervening sources of noise: performance errors, observations errors and other more or less random obstructions.”

But this, says Dennett, need not be the case. “But could the order be there, so visible amidst the noise, if it were not the direct outline of the concrete orderly process in the background: well it could be there thanks to the statistical effect of the very many concrete minutiae producing, as if by a hidden hand, an approximation of the “ideal” order.”

Answer: Many different internal processes could produce roughly the same pattern of behaviors. Very simple processes can produce very complex behaviors.

How is order produced? The bar pattern..

Figure 1. Hard edged process obscured by noise (ten black, ten white, ten black, ten white.)

Figure 4. A normal distribution around black dotes at x = 10, 30, 50, 70, 90.

Then: the serial application of a contrast enhancer. A 3 pixel window is thrown randomly onto the frame; the pixels in the window vote and the majority rules (if the majority are on, then the whole window goes on.). Gradually, black and white areas or bars are resolved.

We can tell apart the two kinds of pattern generators — these internal processes make slightly different patterns — but we may not find that it serves us to do so.

“My point is that even if the evidence is substantial that the discernible pattern is produced by one process rather than another, it can be rational to ignore these differences and use the simplest pattern description as one’s way of organizing data.” Fodor thinks that the “hard edges” of the pattern of human behavior result from the interior representation of propositional attitudes.

In the human case, when we have people who talk, whatever differences there may be the internal processes of individuals, however complex the individual processes that accomplish the multitude of tasks, these differences tend to get blurred by the fact that we are language users. When the different people use the same sentences we interpret them as having the same propositional attitudes.

E.g. If you say “please pass the salt” and a chemist says “please pass the salt” and a four year old says “please pass the salt” we attribute to all three of them the desire to have the salt.

The radical indeterminacy of translation: Intentional attributions to the individual. Ella example.

Two people who wish to interpret Ella: they will attribute two different patterns

with different amounts of noise.

Disagreements are substantial. Don’t always make the same predictions of

individual behavior

a. occasions in which both agree and are right. Yeh they say; we were right.

b. occasions in which both are wrong — this is noise. Boo they say, that was noise.

c. Occasions on which they disagree and one is right and one is wrong. These are merely local victories that may not signal that one pattern is correct and the other false.

Two theories can be reliable and compact predictors but neither, in the long run, is any better than the other.

What his indeterminacy of translation comes to:

“…there could be two different, but equally real patterns discernible in the noisy world. The rival theories would not even agree on which parts of the world were pattern and which were noise, and yet nothing deeper would settle the issue. The choice of pattern would be up to the observer, a matter to be decided on idiosyncratic pragmatic grounds.”

What about Churchland’s eliminative materialism? According to Churchland, we will be best off to throw out propositional attitude talk altogether. Churchland bets that when we find out what is really going on inside, what is happening in brain, our neuroscientific theory will give us far more predictability.

Dennett. Yes, of course it will. But that might mean descending back down to the physical or design levels, both of which will still be much too complicated to use in daily life. OF COURSE, a neuroscientific theory is bound to yield a far more predictive theory, but it may turn out to be completely unusable given its complexity. We will continue to prefer our intentional ascriptions — real patterns with good bit of noise.