The Frame Cursor

A Riff on Kanerva’s Ideas

A Penny for your Thoughts

I posit some small part of the brain, the ‘frame cursor’ that holds some set of bits for a few hundred milliseconds. I suppose that this cursor, “FC” hereinafter, has between 100 and 1000 bits or ‘FC axons’. The FC codes the ‘current activation’ to use a phrase from conventional brain jargon. It is vaguely like a computer’s program counter but I emphasize differences here.

There are several other brain parts able to hold an FC value as well; I will call the Auxiliary FC’s for now. An aux FC is definitely 2nd class to the FC. Aux FC’s have modest axons—only enough to maintain an FC value and deliver it to the FC upon occasion. The aux FCs are what you consult when you ask “Now just why did I look up this word in the dictionary?” or “Why did I go to the garage to get this screwdriver?” or “What else did I come upstairs for?”. Unlike most neurons the aux FC’s are stable for roughly a minute. Sometimes when I have solved some mental problem I close my eyes and try to blank my mind to recover what it was that led me to solve this problem. Occasionally I even write the purpose down before the digression. At other times I look to see what computer editing windows I have open for clues. The collective content of these aux FC’s constitute the extended ‘frame of mind’.

A frame typically includes several FC values which are candidates to be soon promoted to occupy the FC. Which way things go depends on circumstances, of course.

The FC value is a simple compared to what it designates in the brain — a frame. For most FC values there may be a spec neuron, or a few, that fires just when that value inhabits the FC due to how its dendrites connect to the FC axons. Nearly as plausibly the firing of some single stim neuron would cause some specific FC value to enter the FC. There are many (106?) spec neurons but many fewer (105?) stim neurons. There are far fewer (103?) yet FC axons. A stim neuron or spec neuron might behave as mirror neuron.

The part of the brain under control of the FC is very new and not a large portion of the brain. It is the generalized part of the brain that can “think about anything” but just one thing at a time. Other mostly older parts of the brain are special purpose and can do their respective things pretty much regardless of what other parts of the thinking part are doing. I guess that I am adopting the terminology that the FC controls thinking. Within my theory that puts a fairly definite fence around what it is to think. It is a small part of what the brain does. It is the control for Kahneman’s System 2. The thinking part is serial and the rest is parallel.
For those on the lookout for symbols, these FC values would be the primordial symbol. They do not serve that role for they are not accessible to our conscious mind, and they would be useless to transmit between individuals. Ask any programmer why.
A high priority task is to find a way to relate this to tense-mode considerations — Could I, Would I, Should I.

Closely related is the imaginary landmark phenomenon and especially the ‘unconscious’ run-ahead mechanism. This relates to Hawkin’s prediction mechanisms. This run-ahead mechanism may be especially old beginning at least with the vertebrates. There are ‘scenes’ that get stored as the rat traverses the maze and transcribed to long term memory so as to be efficiently and easily accessible when we find ourselves on that path again. A mechanism can run ahead in the head to watch for dangers and opportunities. Indeed this sounds to me like an early selective advantage for largish long term memory. There is no point in asking whether this is conscious for the division has not yet evolved. Its human correlate is unconscious. A physically distinct FC is necessary for this function.

In my rat note I speculate that the transcription from short term memory to long term is contingent on reward. Here I do not.

See note for 1458 here.