28 Feb Sense-Perception and Memory

I Recall…

Since we have been discussing learning in the past few posts, it seems appropriate to talk about memory here as well. Of course, we are talking about human memory, but computers can remember quite well, too. In fact memory is one of the few areas where it is universally agreed that computers are mostly better than humans. Alas, computers are also very good at forgetting. When we completely erase data from a computer, it is gone for good.

Computer recollection in machines of the 1990s arises from three fundamental types of components:

1. Mass Storage (solid state, disks, tapes…) provides long-term memory;
2. Random Access Memory (RAM) provides short-term memory;
3. Read-Only Memory (ROM) provides permanent memory.

As we discuss the mechanics of human memory, we should keep in mind the capabilities of computers. We should try to envision how specific information about the way humans remember will affect the way we organize information for storage in a computer’s memory.

Understanding Context Cross-Reference
Click on these Links to other posts and glossary/bibliography references

Section 4 #22

Table of Context

Remembering

Human memory is extremely robust. Even in cases of extreme cranial trauma, people can usually retain most of their memory. Cases of amnesia are extremely uncommon, and amnesia almost never totally wipes out a person’s cognitive capability. Often the earliest learned cognitive skills, such as communication, are retained, while more transient knowledge is affected. The robustness of memory, particularly in the light of incidents where parts of the brain are destroyed with no apparent loss of memory, has given rise to theories that all human memory or knowledge is totally distributed among all the brain’s cells.

Neural network research has shown that a totally distributed network can retain its “memory” even if some of its nodes become disabled. These results support the idea that it is the distribution of knowledge that makes human memory so robust. How will this affect our model?

We often learn from our mistakes. Why? Perhaps it is because we can see the consequences of our mistakes. If we don’t see the consequences ourselves, there is often someone there to kindly point them out to us. It is the feedback – the more immediate the better – that gives us positive or negative reinforcement for our beliefs, actions and perceptions.

I recall quite clearly the time that a couple of words of dubious origin and meaning found their way into my vocabulary by accident, without my knowing the real meanings. I simply inferred the meanings from the context in which I heard them used. My mistaken use of these words continued until such time as I was kindly, albeit firmly, corrected.

Sense-Perception and Memory

Missouri is the Show-Me State. “Seeing is believing” has been repeated so often it’s become a cliché. According to Aristotle, memory itself requires perception, as does learning. The model of learning and remembering we proposed in our story of Yorrick represents the way the mind works, as described by Aristotle:

“But though sense-perception is innate in all animals, in some the sense-impression comes to persist, in others it does not. So animals in which this persistence does not occur, have either no knowledge at all outside the act of perceiving, or no knowledge of objects of which no impression persists; animals in which it does come into being have perception and can continue to retain the sense-impression in the soul: and when such persistence is frequently repeated a further distinction at once arises between those which out of the persistence of such sense-impressions develop a power of systematizing them and those which do not. So out of sense-perception comes to be what we call memory, and out of frequently repeated memories of the same thing develops experience; for a number of memories constitute a single experience” (Posterior Analytics, Book 2, Chapter 19).

As with a camera, it is the intensity of the impression that governs the clarity of the memory.

Cells that Remember

If the adult human nervous system contains on the order of 100 billion neurons, and some significant portion of these hold information, it is not unreasonable to suggest that billions of cells are dedicated to remembering. Though higher primates have quantitatively similar nervous system structures, human neurons are much more complex. The large number of ribosomes in neurons’ endoplasmic reticulum, or filaments in the cytoskeleton, suggests that each neuron is good for more than just a single binary bit value.

The model we are constructing in Understanding Context assumes that humans have a massively interconnected network of neurons (some 100 trillion connections), each containing complex information. These networks of neurons get stimuli from the biological sensors (senses) and from other neurons through dendrites; transmission of data, or activation, through axons is facilitated by the neuroglia. Assuming an extremely conservative storage capacity of 16 binary bits per neuron, the brain could handle over 300,000 times the data capacity of a 64K node SIMD (single instruction-multiple data) computer like the Connection Machine. If different processes execute simultaneously in the brain, cognition bears little resemblance to the SIMD computing common to many connectionist implementations.

If a single cell can remember a complex data item such as the contours of Gramma’s face, we call it a gnostic cell. It may, however, require many cells to store the gramma data. As we work toward a cybernetic model, we will assume that whether the brain stores knowledge in individual cells, clusters of cells (which is my belief) or the entire brain, we should be able to demonstrate successful brain task models with explicit storage (in specific memory registers) and capture brain processes with models of spreading activation and fuzzy reasoning.

Click below to look in each Understanding Context section

Leave This Blank:Leave This Blank Too:Do Not Change This:

Your email: