The brain and the Memory- Prediction framework theory

The full memory-prediction framework theory was first introduced by Jeff Hawkins in the book On Intelligence. The theory says that the physical arrangement of brain cortex tissue is uniform and means that there is a single principle that defines all brain and memory processing. It also notes that the brain's intelligence comes from the ability to predict future events by relying on past data.

Memory-prediction framework theoryThe memory-prediction framework gives a unified theory about complex behaviors and allows us to better understand what intelligence is.

This theory focuses on the cortex because we're only concerned about what makes us intelligent and not how our entire body works. After all we're trying to create intelligent software, not recreate human beings.

The Central Concept

Memory-prediction framework posits that inputs coming from the bottom of the hierarchy to the top are analyzed and ranked in a hierarchy of recognition. This then invokes a list of expectations ranked in order of potential. The framework comes into play when the brain has to compare and match up these inputs and expectations. The memory-prediction framework means that the brain does not have to consider every option at every level of the process because it uses past sequences as a guide to predict likely future sequences.

"Memory-prediction framework posits that inputs coming from the bottom of the hierarchy to the top are analyzed and ranked in a hierarchy of recognition."

The further up the framework, the longer the past sequences are and consequently the fewer options there are to finish them so the process actually accelerates as it nears the end. For example in looking at a scene, the brain first recognizes lines, then shapes and colors and finally recognizes them as objects. At the same time in the same framework, predictions about what to expect from these objects flows down to speed up our interpretation of the scene.

The framework changes as we age because we add new memories to the system. Since we start with none, as babies we truly see things for the first time. But as we age, we collect memories and this bank of expectations about the way things are and the way things work. This helps us to understand the world around us and to process external information and stimuli more rapidly, but it helps explains how different people can interpret the same situation differently because they bring different memory structures to the experience.

The Pioneer of the Memory-prediction framework theory

Jeff Hawkins originally trained as an electrical engineer and perhaps this gives us a bit of insight on how he approached the problems of brain theory and discovered the memory prediction framework theory. His theory is one of the brain as an organ capable of future predicting and error correction. The brain predicts future events by relying on past data.

"His theory is one of the brain as an organ capable of future predicting and error correction."

The system is a hierarchy so the steps of the analysis are performed in order and if the current even deviates from past experience at any level then a new string of events, or memory, is created. This string can then be used in the analysis of future situations. In this way, the system itself learns and evolves.

It continuously grows more complex and better at making predictions. Just as the theory describes a brain that is constantly adding details and sequences of understanding, the theory itself needs to grow and be fleshed out with details before it will be fully accepted.

Posted in: AI theory, brain and the Memory