Wednesday, 14 March 2012

When tree rings are semantic information

I argued recently that tree rings are natural information, not semantic information. But, suppose I were to control the growing conditions year-on-year in order deliberately to create a specific pattern in the tree rings? They'd be semantic information then, wouldn't they?

The idea comes from this (which I came across via an article in last week's customers' magazine from the supermarket Waitrose):
Barcoding wild Salmon

To increase wild salmon populations in Alaska, hatcheries raise batches of salmon from eggs and release them into the wild. The little fish are held just long enough to remember the scent of the stream where they will later return to spawn.

How do scientists tell the difference between a salmon born in a wild fish hatchery and one born in a stream? The answer, barcodes.

For almost thirty years, fish were tagged by removing an unnecessary fin and placing a metal id tag in the fish’s nose. However, this process was time consuming, and only about a tenth of the fish that came from hatcheries could be marked.

Researchers then discovered that a kidney bean shaped bone in the salmon’s inner ear, called an “otolith,” is sensitive to water temperatures during embryonic development. When the temperature of the water running over the salmon eggs is raised by a few degrees, the developing otolith adds a darker layer of calcium to its surface. This is done repeatedly to the embryos. Later on when the adult fish return to spawn and are harvested, the otolith can be removed and cut in half to reveal a pattern of dark rings.

By reading this “barcode,” the hatchery can tell which batch the fish came from and how long it had been in the wild.

http://indianapublicmedia.org/amomentofscience/barcoding-wild-salmon/
Perhaps the distinction between natural and semantic information is not so clear after all. [Later thoughts after the comments below: I don't think this scenario affects the distinction. It is just that the tree-rings, or the ear-rings, can be used to convey semantic information.]

In a diagram with trapeziums the decoding would be the same, regardless of how the tree/ear rings were created - whether they were due to natural conditions or human coding. The difference would be in the coding. In one case you could draw a trapezium for the encoding, the human control of the growing conditions, in the other, well, you could still draw the trapezium but it would represent the natural conditions.

The point, though, is that in both cases the rings are a medium for communicating information, no different conceptually from, say, a genuine barcode.

4 comments:

sjdalf said...

Natural or environmental information is about the regularity or correlations that exist between two things, whether this correlation is created by nature or is due to human engineering.

So, given your scenario:

"suppose I were to control the growing conditions year-on-year in order deliberately to create a specific pattern in the tree rings? They'd be semantic information then, wouldn't they?"

there would still be environmental information if a correlation existed.

A red traffic light is semantic information in that it is meaningful data. But it also carries the environmental information that the lights on the other road are green for example.

David Chapman said...

Imagine controlling the growth conditions over decades so that I write 'This tree belongs to David' in the tree rings (using, say, ascii coding with wide-ring = 1, narrow ring = 0). That would be semantic information, surely? No different from writing it on label that I staple to the tree?

David Chapman said...

... as always, it comes down to the meaning of meaning? You said that a red traffic light is semantic information because it is meaningful data. Writing 'The tree belongs to David' in the tree rings is meaningful, is it not?

But the tree rings due only to the weather conditions are meaningful too, surely, since they tell me what the weather was like?

Simon D'Alfonso said...

yes, these would be cases of semantic information.

I suppose the point to take from this is that semantic information and environmental information can both be present.