THE ROLE OF PREDICTION IN LEARNING
Based on work by Jeff Hawkins, and taken from his book On Intelligence (Times Books, Henry Hold and Co., NY, 2004)

For generations, we have known students learn “lessons.” Educators are responsible for providing experience after experience in the classroom –we call it teaching, of course. But actually we are orchestrating experiences for a captive audience. In the process, we expect that audience gains new information and skills – so that learners are exposed to, can identify and benefit from their knowledge of existing patterns in life.

Great teachers provide many, varied experiences that are consistent in their respective components. In mastering a formula (pattern) for plane geometry’s area of a parallelogram, we assign a task that requires the covering of a box to house a book; the figuring of the surface area of a wall to paint, etc. It is through the consistency of such sequenced patterns (experience, with objects both abstract and concrete) that memory is planted and thus a repertoire of memories are acquired on which to base predictions about the future. For youth to gain the needed intelligence to cope and flourish in this world, they must have memory of experiences as a foundation to spot patterns; they must use the “expected” patterns to base predictions for maneuvering in new situations. Such patterns become the realities (invariant representations) of the world.

As I extrapolate from Hawkin’s work, we educators need to provide students:
1. Experience. All predictions are based on outcomes and truths (patterns) from experience.
2. Consistency. If there are consistent patterns among the experiences (inputs flowing into the brain), the brain will use them to predict future events.
3. Constant feedback. Brains gain assurance or corrections to realize consistent, correct sequenced patterns to store sound memories for future predictions.

The neo cortex (NC) stores sequences of patterns. As Hawkins notes, all objects in the world are composed of sub-objects that occur consistently together. A sequence is a set of patterns that generally accompany each other, but not always in a fixed order. What is important is that patterns of a sequence follow one another in time, even if not in a fixed order.

The neo cortex recalls patterns auto-associatively, which means that patterns are associated with themselves. Humans see parts of a whole (or familiar, common characteristics) and then recall the complete object or pattern. A sort of “filling in the blanks:” I can read brthd_y even when letters are missing. The neo cortex then stores patterns in an invariant form, which is a generalized form of a sequenced pattern. In other words, we generalize the sensory input from our environment: both concrete and abstract thoughts. We create “summaries” of truth – a picture of a standard “grape” comes to mind when hearing the word, even thought the generalized “invariant form” may be in its totality unlike any single grape you have ever seen. We remember inexactly, without recall precision, the summary of a form or an idea. So the invariant representation is made up of attributes common to the same object. All grapes have a rounded shape, a purple or blush skin, a fragrant odor, etc. This is our truth. The neo-cortex then stores patterns in a hierarch of attributes.

Here comes the big leap. Whereas in the distant past, we equated intelligence with knowing huge quantities of information, and in the recent past we equated intelligence with problem solving – today Hawkins is proposing that intelligence is possessed by those with the ability to predict the future. He advances the idea that prediction is the primary function of the NC, and that the ability to predict is instrumental in developing intelligence. I propose a connection between learning and intelligence. A la Hawkins, there is a sequence of steps leading to learning:
ÿ Appearance of something new (what appears to be a grape, but is a heretofore never seen color and shape)
ÿ Violation of the brain’s prediction of what “should” be, based on prior experience and knowledge of patterns and their sequence/components (memory-driven predictions are incorrect).
ÿ Continual feedback is given in a process, which discovers the violation by measuring the accuracy of the prediction, and then considers new components from reassessing input to spot a new pattern. Then the brain changes the invariant representation in order to re-measure input and to measure a new predictions based on incoming patterns – and if they don’t match, brain makes constant, updated predictions based on new pattern info from sensory input until the prediction is correct.
ÿ The brain has learned!

Intelligence= patterns (instruction) > plant memory (understanding, consistent and correct verified through feedback with student) >memory+ memory combinations>predictions about the future.

Prediction requires a comparison between what is happening and what you expect to happen. It requires feedback in the model, to send info flowing back toward the region that first receives the input. Attention begins when predictions are incorrect. The fruit we have in our hand looks like a grape, but is not purple or blush red. It is much larger, and chartreuse green. So I stare at it, look it over and note its unique attributes, and determine that the qualities that define “grape” are much broader than previously thought. Incorrect predictions result in confusion and prompt this attention. The reaction is not expected, the outcome is not expected, the appearance is out of the ordinary….and on and on. Confusion, a desire to “make sense.” Predictions constantly shift as information and data are updated. So we form a new invariant representation – grapes can be green in color as well as blush red or purple. And they can be elongated as well as round.

We even predict words and conversation we hear. When our visual sequenced patterns that are input contradict the predictions about appearance (seeing an unusual breed of fish, seeing a deformed human) attention is immediately aroused. So the saccade of the eyes feed information to the brain that contradicts the predictions. So it is human nature to stare – to take in the anomaly, to understand the new sequenced patterns of input to form a new memory. Once the new patterns are “routine,” they no longer arouse attention and become part of the overall invariant “correct form.”

Human behavior, therefore, is not rigid. The human neo-cortex directs behavior to satisfy its predictions! Predictions are the essence of understanding. To “know” means one can make predictions about something. If that prediction is wrong, the brain can pay attention and make the new pattern a new memory to adjust the prediction subsequently. New understanding, or learning, results.

Question: how do you deal with and guarantee understanding?
Answer: never be upset by the unexpected – it is how we learn. We must expand the repertoire of patterns to predict the future. Perception (how the world appears to us) does not come solely from our senses. What we perceive is a combo of what we sense and of our brains’ memory-driven predictions. Objects (abstract or concrete) are sequences of patterns that occur together, in a predictable fashion, over time.