The capabilities of the brain far exceed those of any modern day computer.

The primary reason for this difference is the attempt by the computer to only simulate the function of one hemisphere--the left. As we know from the work of many and most importantly Dr. Roger Sperry, the left hemisphere is devoted to the processing of linear sequential information coming from a pattern stream.

The right hemisphere, however, is devoted to the visual, simultaneous processing of images. Though both hemispheres receive input from both eyes, what they do with the information is quite different. The right hemisphere uses memories (stored as pieces connected by axon associations) of images to interpret larger, more complex images. These images are interpreted and the summary is sent from the right hemisphere to the left. This feedback loop of information going back and forth between the two hemispheres, each with their own separate memory structures, is the essence of what makes consciousness and what is ultimately required for success of the Godwhale project.

The unique approach this takes is its attempt to actually ride along with the human brain in its function. Rather than try to create a new evolutionary success, we try to infer and deduce how the brain stores images.

Here are some background facts, in no particular order:

The left hemisphere is devoted to linear sequential processing of pattern stream information. The left hemisphere is mostly gray matter with less white matter. This makes sense when consider how the memory of words is mostly not association intensive. It's intensive of single memory slots. Grey matter is neurons, essentially memory slots.

The right hemisphere is devoted to visual simultaneous processing of pattern stream information but in not just one dimension of feed over time but two or more dimensions of feed over time. The right hemisphere is mostly white matter with only a rind of cortex at the top. This makes sense as we know the brain stores an image not by storing the entire image but through axon-dendrite links with other neurons that represent that piece of the image. Understand: in the brain, there is a neuron that is devoted to seeing a line that runs at an angle of 1 o'clock. Another neuron is devoted to seeing a line at 2 o'clock. A triangle, then, is composed of an 8 o'clock, a 4 o'clock and a 3 o'clock. To remember long term that triangle, the brain stores a link that associates those three other neurons.

After many hundreds of false starts, (as evidenced by the code I will shortly upload) I have come to the conclusion that I need to first create a system that takes temporal visual field inputs, linear numeric inputs, and that the two will be linked in a feedback loop. At first, I will introduce a single point into the visual field. Whatever the system names the point, it will be remembered under that name. Then, it will keep absorbing things it sees for the first time with the goal of having it recognize that two lines at an angle are not just a new object but are, rather, two earlier lines it has seen and memories.

The memory that will take place on both the left (linear sequential) and the right (visual simultaneous) sides will be designed to hold a distinct piece of information: a word, a line, a point, something worth memorizing, and then the associations. The first and distinctive phase will be the learning of these sensory inputs and automatically logging them as learning. The first draft of this project that I published on Google Code used a test dataset that was canned and I now believe that is the wrong approach. It must first learn how to learn--and then how to recognize something it has learned.

Unfortunately, each hemisphere will have a variant of this process and neither can be skipped.

My major decision is to proceed in a Darwinist way, with the success or failure of each iteration being decided on the basis of what it actually learns or predicts.

I welcome any input.

of this is defunct and superceded by what what is above.

This project will design and build the Godwhale, a personal companion based on artificial intelligence, as described in the books "Half Past Human" and "The Godwhale" by TJ Bass. This is a highly ambitious project and it includes several phases that may not be possible at this time, such as:

1. ) Speech Module--this module will convert human speech to XML.
1. Human speech to text--this may not be achievable at this time.
2. Text to XML--the end result of this will be an Idea file that contains a single idea. During the initial phases of the project, a hand-coded copy of this file will represent the starting point.

2.) Thought Module--this module will process the thought-XML file and take actions.

1. Parse the XML
2. Determine keywords, process idea, generate actions.

3.) Action Module--this module will implement actions based on the thoughts.

1. Trigger searches to answer questions.
2. Identify things that need to be pursued.
3. Communicate with the user any outcome from the searches or other actions.

4.) Memory Module--this module will persist any information based on the prior modules.

1. Save keywords.
2. Save associations between keywords. This may be most important of all.

Aspects of this project will include an XML format for modeling thought.