Of Paramecia and Air-horns

Now that we have looked a little bit at how information is (or is not) converted to a signal, it may be useful to revisit the Airhorn Family  and see what they make of all this.

When the air-horn went off, it was simply a loud noise to the baby.  Babies find loud noises scary.  And the rumble of the earth from the blasting was no help either.  But it would be hard to view this information as something encoded in a language the baby could understand.  Or would it?

There must be a reason we find loud noises unsettling.  They make our heart pound and our adrenalin surge.  This can be a great thing when you are at a rock concert but is a very bad thing when you are a sleeping baby. Why would a sudden loud noises cause our heart to pound?  Where is the codebook that says what our brain is to make of this information.  To do anything with it at all, it must be a signal.  And if it is a signal, it must be encoded somehow.  So when did we all sit down and agree that loud noises should be a signal for our heart to beat faster?

To answer this question we must go way back in time and back to the very basic idea of what it means to have senses and to communicate with the outer world.

Suppose we have a single celled organism that lives in slightly salty water.  If the water gets too salty, it  dehydrates and if the water gets too fresh, it basically starves.   Now this little organism lives in salt water lagoons near the outlet of brooks and streams.  The fresh water flowing into the salt water provides the perfect salinity for it to thrive.

The cell uses tiny cilia to "swim" around in the water.  It can basically go forward in a spiral or backward in a spiral.  It would not be a stretch for us to imagine that it could develop the ability to swim back and forth between the saltier water and the fresh, constantly ensuring that the proper salinity is maintained.  Or if you prefer, take 300 of these little guys, with 100 of them always swimming toward the saltier water and 100 always swimming toward the fresh, and 100 of them spending their days swimming back and forth in the sweet spot.  Logic suggests that the 100 back and forth swimmers would find their conditions less hostile than the other two types and so would tend to reproduce more successfully than the others.  Over time, virtually the entire population of single celled swimmers would be the kind that swam away from inhospitable water.

So how would we describe this activity from a signal perspective?  Simply that the salinity of the water signaled to the cell whether it was safe to keep swimming forward or whether it should start swimming backward again.  But how is this signal encoded?  In this case, it is hardwired into the genetics of the cell.  All of the cells that didn't get the genetic memo, so to speak, died.  The ones who survived were all the ones who properly encoded salinity as a signal for how to move.

It doesn't matter whether the cell is conscious of the choices it is making.  All that matters is that, like the resistor that dutifully takes a current and impedes it, the paramecium responds to salinity in a consistent way, swimming forward for as long as the water is safe, backwards as soon as it is not.

This example could be extended quite a bit through many thousands of generations to a level of navigational sophistication that extended beyond our example, but we don't need to go any further to demonstrate the basic point.  On a very crude level, information can be converted into signals at the genetic level.

As Darwin so brilliantly pointed out, the basic method for "teaching" any gene pool how to behave is to simply kill off those who are behaving badly.  Only those who prosper will successfully reproduce (at least in large enough numbers to bother with), so over time genetic selection favors encoding that matches the environment.

Of course we all know this as "survival of the fittest", but this phrase has taken on a bastardization, especially in the social sciences.  To some the phrase means that only the grade A, prime cut members of a group survive -- that the weak ones all die horrible deaths in the big bad jungle.  But Darwin did not mean fit in this sense.  His expression was intended more as "survival of the most apt".  That is, the mechanism of evolution he described was one whereby the creatures who lived would be the ones who had properties and behaviors most suited to their environments.  To take our paramecia as an example, there may have been one really buff paramecium -- the envy of all the others and the starting QB on his high school team -- but if he had only the ability to swim toward fresh water without regard for its salinity, there will be no buff juniors running around any time soon.  His "fitness" does not matter.  His "aptness" (or aptitude) does.

So let's bring this on back to the baby human.  Is there any reason we can think of that sudden loud noises would be genetically encoded to cause a baby to cry?  Of course there is.  A sudden noise in most any natural context for most any creature we can think of would signal danger.  It could be the breaking of a branch to a bird or the sudden pounce of lion to a gazelle, but in most natural contexts, a sudden sound has a high likelihood of needing a quick response.

This means that those creatures who responded to sudden noises with a shot of adrenalin and increased heart rate for a boost of speed and strength would tend to out-live those who had a more laid back approach to sudden danger.  The fight or flight response is simply a genetic certainty.

Now, we could have easily said that the baby was predisposed to not like loud noises and that was why he cried.  But our little thought experiment has taken us beyond that.  Now we can suggest with a straight face that the baby human is genetically encoded to fire up his fight or flight response upon hearing the signal of a loud noise.  So why the crying?  Well that is the baby's response to all the built up energy.  His body just jolted him with energy.  That doesn't mean he is suddenly blessed with adult coordination and problem solving, however.  Those will come later.  For now, the jolt simply cries out for release.  The energy is burned off with a good cry.

I said previously that the baby had no context in which to place the information of the air horn and loud blast.  But that is not technically true.  His conscious brain did not have a context built yet, but on a much more basic level, the context was clear.  Loud noise signals danger.  In this case, the baby was not far off (even though the danger was not to him).  But it also explains why any loud noise, even a happy loud noise like celebratory party favors, may cause a baby to cry.  To know that a given sound is a good loud noise and not a dangerous loud noise takes context beyond the baby's capacity.

Now the cat is very much in the shoes of the baby, context-wise.  In our earlier piece we granted her some perspective she may or may not have had.  My assumption is that animals are generally smarter and more aware than we give them credit for, so I see no reason the cat couldn't learn some predictive elements about the noise -- such as that it never happened at night.  But in any case, the response was very similar to the baby's:  it fired up the cat's fight or flight response.  Again, this is because this signal was interpreted by a very deep part of the brain, encoded with ancient signals stretching back to the dawn of critters.  But the cat had the means to do something with all that adrenalin and she used it to run and hide behind the couch (seeing no obvious opponent to "fight", it was "flight" time).

When we treat information as something that becomes a signal when put into context, we realize more than ever the power and significance of what context is.  For even the baby who was still sizing up the world around him came pre-packaged with a context with which to view the world.  This would, in this case, be the context that is genetically encoded in his mind, his "instinct" for lack of a better phrase.  Only the Mother and Father had the experience and capacity to place the air-horn into a more sophisticated context. And there again, we should remember only the construction worker actually heeded the signal in its intended form.  That is, whatever else the air-horn meant or symbolized to him, when he was on the job site he took it "literally" -- as a warning that a blast was going to occur.   When the air-horn blast served to remind the construction worker's wife that he was at work nearby, it was still a signal.  But it was a signal in a context that she had created herself.  Her context was the result of her relationship to her husband and all of the significance and stress that she assigned to his job and career.  She could "hear him at work", which she found warmly reassuring.  But she was also reminded (not necessarily consciously all the time) that his job was sometimes dangerous.  She knew the air-horn could wake the baby. She knew it scared the cat.  She knew it was sometimes just an annoying distraction from what was on her mind.  The air-horn was all of these things to her.  And each of those meanings was associated with a context.  So for her perhaps more than any of the others, the information of an air-horn blast was a large number of signals.  Her mind quickly parsed each signal every time she heard the blast.  Depending on the facts of her environment (e.g. whether the baby was awake already or had just been put down to nap), her response to this signal would change.  But each time the information was received, her various contexts all played a role in deciphering the signal.

Red Means Stop

Information vs. Signals

I have talked about information a lot without ever defining it.  I have treated it as the implicit data that is present in the outside world.  The coffee is brown, it is 145 degrees, it contains 140 milligrams of caffeine, etc.  Technically the color brown is a perception, not a piece of data, but we can use it as shorthand for “this coffee reflects light in a combination of wavelengths which we call brown”.   But there are a lot of schools of thought about what is actually meant by “information”.  For example, can a message contain information if it has no meaning to the person receiving it?   It depends on who you ask and what you mean by “information”.

Claude Shannon, a pioneer (or THE pioneer) of “information theory” thought of information as nothing more or less than the surprise disorder in a stream of homogenous or predictable data, whether or not this information had any “meaning” at all.  In other words, information related to Entropy (the tendency towards disorder).

Entropy can be briefly described as the tendency of the universe to move toward lower energy states and higher disorder.  For analogy’s sake, consider two tea cups.  One is on a shelf completely intact and the other is on the floor smashed into 100 pieces.   Entropy says that it is much more likely for the tea cup on the shelf to join its shattered companion by falling off the shelf and breaking (achieving both a lower energy state and higher disorder), than it is for the broken tea cup to suddenly assemble itself and jump up on to the shelf (achieving a higher energy state and greater state of order in the process).

If Entropy is the tendency towards disorder, information conveys the breaks in an orderly state – the disorder that we can find within predictable data.  For example, a white piece of paper could be described as having very little information.  It is white.  That is all there is to it.  It is full of order and sameness.  But if it were full of random letters in a variety of colors the paper would contain a lot more information – a lot more disorder – even if the letters and colors had no symbolic significance (no meaning) to us.

This way of thinking about information arose from thinking about how to transmit data about the world (through telegraphs, telephones, television, etc.).  And seen in this way, we can kind of get what is meant in this case by “information”.  It would take very few pieces of data to describe (or transmit) our white piece of paper.  We could for example say “white paper 8.5x11 inches, blank”.   But imagine how many characters it would take for us to describe (or transmit a copy of) the paper if the paper were full of multicolored letters as we have said.  Nothing short of describing each letter and color would suffice.   This inability to compress the description of the object – the need to devote more transmission elements to accurately portraying it – means the multi-colored paper full of characters contains more “information” than the blank page (At least using this definition of information).

There is a halfway point between the blank page and the random letters.  Supposed the alphabet was repeated on the page first in Red, then Orange, then Yellow, then Green then Blue then Indigo and finally Violet.   It would be harder to describe then the blank page but easier to describe than the random letters and colors.  “White Paper, 8.5x11 inches, English Alphabet repeated, each copy in one color of the rainbow starting with Red and going in spectral order.”  There is, in this case more information than the blank page but less than the random letters and colors.  This is because the data is ordered.  The higher the degree of order, the less information it contains and the less data is required to describe or transmit it.  This is why Shannon described information as being the “surprise” in the data.  Very ordered data with few surprises contains very little “information”.  It makes no difference whether the data spell anything or stand for anything or have any significance at all.  All that matters is how they are ordered.  In short, there is no connection between information (in this sense) and meaning.

So what are we to do if we wish to discuss information in terms of meaning?  Well one handy method is to use a different word altogether.  In this case the word I will be using is “signal”.

A "signal" is any piece of information which carries symbolic meaning for its receiver.  Or, more generally so we can include non-thinking signal processors:

A signal is information which has an encoded implication in a defined context.

Consider for a minute that you are walking along in the woods in the evening.  Off in the distance you spot a church steeple.  Two lanterns hang there, burning brightly.  To you this is information, not a signal.  But if you are Paul Revere, and it is April 18^th, 1775, the light is both information AND a signal.  It means the British are sailing up the Charles River.

What takes information and turns it into a signal?  The language or code you choose to unlock its meaning.  This code need not be spoken language or written language or anything nearly so sophisticated.   The information must simply be encoded somehow.  There merely needs to be an agreement or policy that certain information STANDS FOR SOMETHING.  To every American who knows how to drive, or knows anything about traffic at all, a red traffic light is a signal to stop.  Traffic lights are signals which are so well understood by everyone that they are even referred to in some cases as “traffic signals”, or even “stop lights”.   But if you were an alien who had never seen a car and did not know anything about traffic, or even if you were simply Paul Revere pulled forward in time, you would not gather any meaning at all from the pretty colored lights hanging over the roadway.  They would still be “information” but they would not be a “signal”.

So what is the process that absorbs information and turns it into a signal which we can comprehend?  It is quite simply that -- a process for signals.  We’ll call it “signal processing”.

Signal processing takes place any time there is information which is translated into a signal.  This can apply to anything from the simplest electrical component to the most highly ordered brain.  Information is constantly being absorbed and some percentage of that information is translated into signals which carry (at least at the cognitive level) meaning.  At the lowest level of signal processing, we may still use the term “meaning” but there is no cognitive understanding taking place.  (Cognitive understanding and consciousness will, not surprisingly, be set aside for another time.)  Even if there is no understanding though, there is still a discreet action triggered by the signal, so when we say, the motion detector senses motion and that “means” it should turn on the light” we are being philosophically sloppy even while we are being mechanically precise.  The difference between a signal actually carrying meaning and merely triggering a mechanical response is a subtle one full of all kinds of conjecture and puzzles.  It is surely worth coming back to.  But for now we will simply say that a signal X “means” Y.  Two lanterns “mean” the British are coming up the river (if you know the code) and the motion in the yard “means” the light should be turned on (if you are a motion detector which has been programmed properly).

Signal Processing is Transforming Information

Let’s take two cases of signal processing side by side – a very simple one and more “high order” case.  The simple signal processor will be the modest electrical component called a resistor.  The higher order signal processor will be a celebrity’s assistant.  We can see how each of these signal processors function in similar ways and transform the information they receive.

In the case of the resistor, an electrical current comes into the base of the resistor.  The resistor provides drag or “resistance” to the current and some lesser amount of current flows out the other end.  The sole purpose of a resistor is to reduce the amount of current flowing in an electrical circuit. 

One of the duties of the celebrity assistant, on the other hand, is to read the celebrity’s mail (at least in this example).  The assistant reads all the incoming fan mail, business offers and junk mail and filters out most of it so the celebrity does not have to waste her time reading it.  Then when a particularly interesting piece of fan mail or a business offer or personal correspondence comes through the data stream, it is passed on to the celebrity for reading.   In this case, the assistant acts very much like the resistor.  He takes a heavy flow of mail and reduces it, passing on some lesser amount of letters through the other end in a fashion very similar to how a resistor reduces current.  Both components – the assistant and the resistor – are engaged in signal processing.  They take some information and transform it, passing it on in a form that is more useful to the system they are working within.

Now of course, the degree of signal processing that must take place for the assistant to do his job is much more complex than that of the resistor.  The resistor simply provides drag, but the assistant must collect, open and read all the mail. Reading the mail requires a great deal of higher brain function and higher still is the function that allows him to assess the pieces of mail for content and assign some level of priority to them so his employer will only see the important pieces.  This is actually a very long chain of complex signal processing all wrapped up into the task we call “screening the mail”.  But the basic concept is the same.  Some information comes in which is treated as a signal.  Those signals are transformed and pass out the other end as new signals which are new, more useful, filtered, or whatever characterization you want to use.

Let’s revisit the traffic light.  I am driving down the road.  I approach a light in the distance.  I see that it has turned red.  I perceive this information and it is translated into a signal in my mind. The signal says, “stop the car”.  So that signal is then translated into action – I downshift and step on the brake, bringing the car to a timely halt.

See what has happened in the course of this signal processing?  There was Information (the red light) which I put into Context (translated into a signal) and then I took Action (the signal inspired the complex motion of muscles that resulted in the braking and downshifting associated with the skill I had previously learned called “driving a car”.)   What do you suppose would happen if I was hiking through the woods and hanging from a tree was a traffic signal that displayed a red light?   Would I stop and wait for the light to change?  Not likely.  Why not?  Because signals have no meaning when they are out of context.  That is why I have spent so much time considering the role of context in information.   The information is the same as ever – a red traffic light – but the context – walking in the woods – robs the signal of any meaning.  I have no car to stop even in the unlikely event I was convinced that’s what it was telling me to do.

As a fun aside, consider that I would probably retell the story to my friends when I returned.  And I would probably use the word “disorienting” when I recounted how I spotted the traffic light hanging from a tree.  When we are met with information that implies a context but our current circumstances don’t allow us to construct that context, we are “disoriented”.  The word itself spells out how much we expect to be aligned to a certain frame of reference – a context – in our daily lives.  Seeing someone from work when we are at the grocery store can be disorienting, because we want to place that person into the context we call work, but we are not at work.  We will consider much more about the effects and implications of being disoriented another time.

So every signal, to be a signal, must be associated with a context.  It is part of the coding process for the signal itself.  So a traffic light means certain things when I am driving a car down the road.  It does not mean anything to me hanging on a tree in the woods or from the wall of a restaurant or any other place outside of the encoded context.   Suppose Paul Revere had come across two lanterns hanging from a tree in the woods.  Do you think he would have started his famous ride convinced the British were coming?  Of course not.  The lanterns meant something, but the code only applied if they were hanging in the steeple of the Old North Church.

In the real world, contexts overlap.   I know that a ringing sound coming from my phone means I am getting a call.   But the same ringing sound coming from the television does not mean I need to answer my phone.  That assumes of course that I can tell the sound is coming from my television and not my phone.  Anyone who has ever reached for his cell phone upon hearing someone on TV get a call was engaged in very complex signal processing.  The information of the sound was perceived.  The signal was interpreted within the context of the cell phone. The context of watching television was also present but was temporarily dismissed.  Until, of course, the blank screen on the phone makes it clear that no one is calling.  Then the proper context of television is applied to the information and the right action (doing nothing) is applied.

The fact that contexts can overlap and are not always clear greatly impacts the outcome of our daily signal processing.  What’s worse is that some contexts can conflict with one another.  Working at your job and needing to finish a task can compete directly with needing to leave work for a family emergency.  And this is just a very basic conflict of contexts.  Much more subtle conflicts exist constantly.  You may be interested in impressing a young woman at work and engaged in clever banter but you also know that you have a lot of work to do.  You legitimately want to be successful at each task – the work and the socializing – and the overlap of the two contexts is clear to you.  That does not make the signal processing any simpler.  The signals you receive and the things you say and do every day are the result of dozens or even hundreds of overlapping contexts.  Choosing when to get up and use the bathroom should be a basic biological action we are well equipped to perform.  But when the information that your bladder is full conflicts with the expectation of those at the meeting that you will hear what they have to say – or perhaps your own desire to watch and see what happens on screen in the movie theater – simple Information-Context-Action scripts do not unfold predictably.   Just because the process can be described in simplified terms is not meant to imply that the process is itself a simple one.

But let’s look more closely at the process of gathering information and placing it into context.

PERCEPTION

It is sometimes assumed that the act of absorbing information, commonly called “perception”, implies that information acted upon in the brain corresponds with what exists in the “real world”.   In fact, though, the act of perceiving is the first step in context formation.  Just as chewing food is the first step in digestion, perceiving information is the first stage of signal processing.  There is no such thing as solid information hitting the brain for signal processing (context formation) any more than there is whole unchewed food sitting in your stomach awaiting digestion.  In order to swallow the food it needs to be transformed into something more manageable.   So it is with information.  In order to absorb information, it needs to be transformed on the spot by our sensory organs into something our minds can “swallow”.

Sometimes this transformation of information can be problematic.  If you have ever jumped at a shadow you know what I mean.  The eyes did not see a shadow, dutifully report to the mind that it was a shadow and leave the mind to decide “yes, but it is a scary shadow. I should jump.”  On the contrary, your eyes passed along something that was NOT a shadow but something solid and moving and very near you.  You responded not to a shadow but to what you were told the shadow was.  Once you discovered the shadow was a shadow, you knew your response was out of proportion.  Once your eyes corrected their mistake, you could respond more normally.

But hold on here.  Aren’t the light waves just light waves?  Isn’t all the information contained in the lightwaves simply passed on through the optic nerve?  Isn’t it the brain’s fault if something is misinterpreted?

This could easily break down into semantics, but there is no logical reason it should.  If you think of “seeing” as including all the sensory elements involved in collecting visual information, the “seeing” isn’t done until the light waves have been transformed into the image that the brain can work with – just like the food isn’t swallowed until it has been chewed.   So whatever mental processing takes place to interpret the form and its position in external space is rightly categorized as part of the perception process.  When I said that perception is this first step in context formation this is what I meant.   You haven’t perceived an image until your mind has been told what that image is (not what it "represents", but what it is made out to be – a solid branch on a tree bouncing in the wind for example).  The recognition that the branch is a branch is part of the act of seeing and is part of the perception process and, as stated, the first step of context formation.  So if the branch is really a shadow but your mind is told that it is a branch, the information processing performed will be tainted by the mistake your eyes have made.  Your reaction to the branch might be perfectly normal (you may duck).  Only the fact that it is a shadow and not a branch makes your behavior at odds with reality.  But you still reacted appropriately to the information you were given.  You were just given bad intel, that’s all.

Why does this matter?  What is the benefit of breaking out these processes in such a way?  Well simply put, this model will help us understand any number of seemingly unwarranted reactions as actually very normal and appropriate reactions that have been placed in the wrong context.  If we want to look at signal processing, we need to be clear how the basic machine works in order to diagnose its many malfunctions.  So perception should be seen as that initial stage of information absorption.  But it must also be viewed as the very first step in the process of context building.

What is seeing really?  Can we declare that there is a dividing line in the brain that really separates vision from thinking?  As usual, our answer is yes and no.  Consider a coastline on the sea.  We can tell from a distance that there is ocean and there is land.  The distinction is not a controversial one.  Some creatures living in the ocean can not live on land and some creatures living on land can not live in the sea, so there surely must be a difference.  But if we walk along the beach, can we say with any certainty where the ocean ends and the land begins?  With the ocean to our left and dry land to our right we can point in either direction and say, “that is sea” and “that is land”.  But when we look down we see the waves crashing at our feet.  Advancing water swells and retreats, sometimes leaving little pools of water behind.    Where does the sea start and the land end?

This problem is very similar to the types of challenges we face in drawing boundaries in the mind.  There is always a grey area about which function or process falls where.  But just as with the land/sea dilemma this does not mean the two regions do not exist separately.  It only means we can’t always tell where the edges are.  Of course, this is the same problem we had with truth and falseness.  And the same advice applies.  Just because the border is grey does not mean they aren’t real and separate things.

It is not trivial to realize that we face the exact same quandary when trying to draw a line between the “in here” and the “out there”, that is, our “selves” and our “environment”.  We exist in the world, we are part of it, but we are also separate from it.  There is a vast cosmos of internal space that is my mind which I can not share with the outside world.  And there is a vast world of things and people out there which function every day without my knowledge.  (Consider for example the number of people on the planet who are at this very moment waking up from a night’s sleep.   I do not know them, I can’t count them, and I can’t imagine them all as individuals. Yet they are real people living their very real lives, completely separate from my own human experience.)

So the universe, whether mine or yours, contains an out-there and an in-here.  There is no doubt about the distinction.  But that does not mean that drawing the line between the two regions is any easier than determining where the land becomes the sea.   One thing we can say with some confidence, however, is that our perceptual systems occupy the sandy beach separating the inner mind from the outer world.