[Return to Index page]

Charles Darwin appreciated the creative power of natural selection.

It was evident in paleontology - the fossil history of life on Earth, and in his observations of the divergence of contemporary living species.

Today we see natural selection in many places besides genetic evolution.

In our immune system antibody molecules are produced by cells in our blood stream that are self-selected to proliferate, by their reaction to foreign body antigen molecules. Gerald Edelman got his Nobel prize for his work on the biochemistry of this adaptation. 

In anatomic development, especially in the nervous system, cells proliferate and then die unless they make the connections to become functional. Natural selection again.

Neural networks adapt by natural selection. There are about 100 billion neurons in the central nervous system, each making connection to about a thousand other neurons.  After early childhood nerve cells ordinarily do not reproduce, but the interconnections, the synapses which can either stimulate or inhibit a response, are reinforced or attenuated depending on their activity.  By this means clusters of nerve cells become specialized according to the inputs they receive, from sensory organs, from surrounding or distant neurons, and from positive or negative feedback from near and far in the brain.  Thus they are constantly combing their connections for combinations that are significant, and reinforcing the ones that turn out to be useful.

Every neuron is a little decision-maker.  The cell membrane acts like a chemical battery, maintaining a several-millivolt potential difference between inside and outside.  When the neuron is stimulated, by an appropriate combination of input signals, a process occurs that allows potassium ions and sodium ions to penetrate the cell wall, which is like short-circuiting the battery.  This produces an impulse of millivolts, that lasts several milliseconds, which propagates along the nerve fiber at hundreds of feet per second.  Then the system recovers to repeat a few hundred times a second.  Each impulse becomes part of the combination of inputs to each of  the thousand other neurons that might be stimulated in association with the first.

So we, and other animals, have evolved the associative capability for pattern recognition, which becomes the basis for memory, and imagination, because related inputs can restimulate patterns from previous experience.  [ * See paragraph added 28 June 08 below * ]

For millennia philosophers have been mystified by what they call the mind-brain problem.

They could theorize about the character of real things, and their interaction, and they could describe their inner experiences but what kind of reality was that?  Our perception of ourselves and of our surroundings is not a "physical thing" but it obviously exists, so it must exist in another domain, call it "spirit."

But these separate domains obviously interact. We perceive things and we manipulate them. In fact our body is a thing and we trigger its muscles to manipulate it, and tools, and other things. We even perceive a tool as an extension of ourselves, and feel the force of the pliers on the object we are gripping, or the impact of the tennis racquet on the ball. Yet the perception is obviously different from the physical fact. How does that work?

Today we have clarified our thoughts about information and how we use it to wend our way and work our will in the world.  Norbert Wiener, in his 1948 book Cybernetics: Control and Communication in the Animal and the Machine was fully aware of the broad significance of his new science to provide a vocabulary of ideas to conceptualize these aspects of real systems.  The name Cybernetics is coined from a Greek word for governor or steersman.

Experience and purpose and behavior are ideas that we use to comprehend what happens.

The ventilating system in this room has a very simple central nervous system.  The thermostat processes one bit of information - one binary digit that signals when the temperature is above its set point.  That signal trips a relay that turns on a cooling system and then turns it off when the room temperature is below the set point and the signal disappears.   We understand the working of that system as a process - the rudimentary experience of the thermostat results in behavior of the system to regulate room temperature by using an external source of energy. That system was created with a purpose - by its Intelligent Designer - it's self-motivated to keep the room comfortable.

What was mystery to Plato and Aristotle, and Descartes, and Kant, and Whitehead is now, more and more, empirical fact. What was speculation to William James and Donald Hebb has become neuroanatomy and physiology.  The same functionality that facilitates the habitual repetition of familiar actions without conscious thought, the practice makes perfect of athletic performance, and which was studied in Pavlov's conditioned reflex experiments, is now being exposed in all its details.

Take for example our system of visual perception. In the 1950s it was established empirically, with surgically placed electrodes in cats' brains, that fibers in the optic nerve don't carry point-by-point signals from the optical image projected on the retina, but rather are stimulated by abstract features of that image. The retina of the eye is part of the central nervous system, with associative neurons that connect to a variety of combinations of the surrounding rod and cone cells that are sensitive to light. They detect patterns of signals that represent features like edge contrast and geometric orientation and brightness or color. The impulses through the optic nerves to the brain represent patterns, not points in the visual field.

In the brain there are at least seven regions that receive signals, directly or indirectly, from the optic nerves. They specialize in different aspects of pattern recognition - object vs background, motion, perspective, hue and shade of surfaces corrected for differences in illumination, and so forth. They are triggered by patterns of patterns and they feed into one another to create the images in our brain that represent our surroundings. I'll try to show you an intuitive idea of how that works. 


What does pattern recognition accomplish?

Each associative neuron is constantly combing its inputs for a combination that it recognizes, and it says "Aha!" by sending a signal to the multiple neurons that connect to its output.  They receive this input and many others, and generate signals to other neurons.  And some of them say "Aha - I recognize that combination of edges and surfaces - it might be a table." And that output triggers other neurons that say "If it is, then it must have certain characteristics." And other neurons that say "If so, it is. If not, it must be something else."  Of course all this is non-verbal processing and we use word analogies to describe it.  This roundabout way of discovering reality works so well because each associative neuron is self-motivated, checking its fan-in of inputs and waving its fan-out of outputs, working in parallel with billions of others.  And when a combination of stimulative and inhibitory signals causes a useful output that combination  is reinforced, and can be more easily recalled.  Thus we build up a memory of past experiences that feed back into our recognition of present sensory inputs in what Edelman calls a re-entrant system.   The associative character of that information storage and retrieval is natural for a parallel - processing system of neurons that are each doing their thing, and so much is happening that we can only be aware of the naturally selected, significant results.

Let's realize that our perception is a fantastic process of pattern recognition - fantastic because we can indulge the fantasy of what might be, related to what we perceive. We automatically fill in the blanks in our visual field, the blind spot where the optic nerve exits the eye and where blood vessels cross over the retina. (The retina of the eye is anatomically inside out! The light-sensitive cells are on the outside and the connecting nerves and blood vessels are inside toward the lens!) We use our knowledge from past experience of familiar objects to recognize the whole object, not just the part that is visible from a particular viewpoint, and we correct the image automatically for the motion of our eyes. We see in our mind what's there, and what it's doing, not just the fragmentary optical image in our eyes at any moment.  

So all we know of an experience is the thoughts it stimulates.  Therefore recalling the thoughts recalls the experience.  But the thoughts are information and the experience is its display in our mind - rerunning the associations that give access to their whole context.  As Edelman suggested, our experience of reality is our imagined present. We are modeling our surroundings in our brain.

A model is a representation of something else.

It's a construction that incorporates information about something that it represents.

A ship model is a miniature construction that communicates a reality that is too big to fit on the shelf. A dress model is an idealized wearer of the product. A model of behavior is an example that embodies an ideal.

And in science we recognize that our ideas about real things and processes are idealized representations - they are models of reality.  But that's another whole lecture.

Another example of modeling will clarify our thoughts about reality, and the reality of our thoughts.

Architects don't do architectural drawing any more, except for casual sketches to develop their ideas. They use computer programs that compile lists of data about their creation - locations of points and lines and surfaces in the construction. That input is facilitated by a user interface - a part of the program that displays on the screen an image of what the data represents. When the job is done the computer uses the data-list model to drive a plotter to create working drawings to represent the project, another kind of model, with dimensions and specifications for construction. And perspective drawings with shading and rendering of surfaces and surroundings to visualize the completed project.

A related application is called virtual reality - a program that accesses the data about a project and creates an image on the screen of what it looks like from any viewpoint, inside or out. We can walk through the building and see it as it will be when finished.

An even more impressive example of virtual reality is used in flight simulators and video games. The user wears goggles with binocular images, which are rigged to signal the computer about the user's head position. The resulting images change automatically as the user looks around at the virtual reality represented by the computer model. The result is a very realistic experience of "being there."

But the map is not the territory, as the General Semantics people tell us.  It's a model that displays information about the territory, incorporated in a physical form of paper and ink.  Or in the GPS system you might have in your car, that calculates your location from satellite signals and refers to map data to create an image of the roads you might take from where you are to where you want to go.

So we see that this visual modeling process has three components - data to represent the reality, an implementation to view the data, and a user to react to what he sees.  In our mind the data is sensory signals and the associative recall of relevant experiences, the  implementation is the selective presentation to our conscious awareness of the sensory inputs in the context of our knowledge about their relevance.  The user is our stream-of-consciousness awareness function which automatically selects associative pattern-recognition results that are relevant to our needs.  We have evolved this selective attention function because so much goes on in our mind that we can only deal with a few refined results.

The modeling function in our mind serves a very powerful purpose - Extrapolation!  From sensory inputs and past experience we can imagine what's coming. We know where to go to catch that fly ball when it comes down. How do we know? The associative connections in our massively-parallel-processing system call back aspects of past experiences and fit them together into our ongoing imagination of what-to-expect.  We know that that fly ball will continue on its trajectory unless something gets in its way.  And that experience is dynamic because it is constantly updated by new sensory input as we zero in on our effective motions to catch the ball.

We automatically forget what was wrong in our perception of reality and replace it with the latest information. We, and other life forms, have evolved the scientific method!  For its survival value! We theorize (imagine) what to expect, compare with empirical data, and correct the theory so that it better models reality. A bat in the dark, flying by sonar, or the common fly, with very different optics, can approach and land wherever they want to, using similar but more rudimentary versions of awareness of their surroundings.

We have used visual perception here as an example because the abstract character of associative pattern recognition and retrieval is so well-known and understandable as data, display, and perception/reaction.  But this parallel processing pattern applies to other brain functions, such as kinesthetic motion control, speech articulation and recognition, and abstract reasoning.  But that's another whole discussion of long thoughts about reality.

[ * Paragraph added 28 June 08 * ]  Incidentally, this amounts to content-addressable memory, an idea that has been developed in a rudimentary way for sequential-processing electronic computers.  They use memory locations that passively store multiple bits of information that have meaning in context, and hardware to decode location data and trigger the appropriate location to output its content.  Specialized hardware can broadcast selected data, for example key words of a record in a gigantic database, for comparison with query specifications and assistance in locating the record.  But multiple index files work better and don't require specialized hardware.  With parallel processing by active memory elements that can themselves recognize a match, content-addressable memory is natural.  [ ** ]

So we see how the parallel processing with billions of associative pattern recognition elements yields the fantastic wealth of possibilities in our thoughts about experiences, real or imagined.  The magic of our mind is a natural product of the proliferation of self-motivated associative connections, and the natural selection of those that are relevant.  Spontaneity from the proliferation of possibilities, and purpose from the context of selective attention.  And the magic of intuitive leaps to new ideas that we create.  And the creativity to play GOD, whether we want to or not.


Here are some references which have been important to me:

I make no claim of originality, only a continued interest in the subject.  I see these ideas as a consensus and a work-in-progress among investigators in the field of cognitive psychology.

Second Nature - brain science and human knowledge.   Gerald M. Edelman, Yale University Press 2006.  A brain-based approach to consciousness.
Bright Air, Brilliant Fire, Gerald M. Edelman, Basic Books (Harper Collins) 1992.  Subtitled On the Matter of the Mind, this popular development of mind as a biological process uses the Nobel laureate author's command of developmental biology to explain the functions of awareness and intention from an evolutionary viewpoint.  He uses the term "the remembered present" to characterize our internalized idea of our surroundings.

Patterns, Thinking, and Cognition - A Theory of Judgement.  Howard Margolis, U.of Chicago Press, 1987. An evolutionary approach that emphasizes the dynamic balance between reconsideration and commitment to action, that is observed in "lower" life forms as well as mankind.  Margolis shows that pattern recognition is an iterative process of successive approximation between an internal representation and external stimuli, and that imagination and invention are part of the process. He applies these insights to societal behavior, in particular to the century-long interval for the general acceptance of the Copernican paradigm for our solar system, first by navigators and map-makers and last by professional astronomers.

The Emotion Machine, Marvin Minsky, Simon and Schuster 2006.  A proliferation of "Ways of Thinking" that articulate the diverse ways the mind works.
The Society of Mind, Marvin Minsky, Simon and Shuster 1988.  An Erector Set of essays about the mind, as an interconnected system of self-motivated agents.

Norbert Wiener, Cybernetics: Control and Communication in the Animal and the Machine,  MIT Press 1948.  The classic introductory work on information theory.

Helga Kolb, Eduardo Fernandez and Ralph Nelson,  WEBVISION - The Organization of the Retina and Visual System  A very detailed, state-of-the-art, exposition of the neuroanatomy and functional physiology of the retina.  On the Internet at:  http://webvision.med.utah.edu/

Cognitive Science Society maintains a web site with links to societies and researchers in many related subspecialties.   http://cognitivesciencesociety.org/index.html