Print this page

Common steps and sub-activities

Learning - synthesis

In the process of learning, Perceptions are first extracted and then ‘tagged’ in order to understand what we have perceived.

The next stage is to try to deduce if there is any pattern in the tagged results, or to put it another way, is there a System?

This pattern matching is a form of logical classification, influenced by Objectives, Threats and so on plus the culture and people round us. We are thus building up belief systems about how types of thing inter operate. We use the tagged occurrences and try to build up a picture of how first these can be classified [the Database of facts] and secondly if there is a recognised system in how they are behaving [Learnt function].

Whether the process of synthesis takes place during dreams or during waking life, the process seems to be similar.

The first thing we seem to do is see if the new Perceptions fit an existing pattern, in effect the new observations we have received are simply a reinforcement of that pattern. If the existing Database of facts has classes and relationships which all seem to fit the observations we have, then all that is required is for the data – the new tagged occurrences to be added to the database.

Next, if they don’t match, we may then look to see if an extension of an existing model can handle the new data – whether this is new classes and relationships in the database or new activities and activity dependencies in the learnt function model.

Only if the observations cannot be fitted anywhere do we start to build new models with new classes and new learnt functions. The tendency always seem to be to want to build on what we have, rather than start something new.

Dreams are used to learn. There is also evidence that during dreaming the extracted Perceptions are compared to our existing Database of facts and model of functions, to see if the new data can be slotted in or extensions need to be built

In order to demonstrate the process, I will now return to the simple example I used in the section on Understanding and recognition …..

Database of facts 

The Perceptions on the left broken down into a series of ‘statements’ of activity – each themselves preserved in the sequence they occurred were, in the last process of learning, given a ‘tag’ to help in understanding what was going on. So on the right, we have the tagged activities also in sequence as before

OCCURRENCES

 

TAGGED OCCURRENCES

Thing A verb Thing B

 ——→

Person A goes into Shop B

Thing A verb Thing C

 ——→

Person A talks to Person C

Thing C verb Thing E Thing D

 ——→

Person C takes Chocolate Bar E from Shelf D

Thing A verb Thing C Thing F

 ——→

Person A gives Person C Money F

Thing C verb Thing A Thing E

 ——→

Person C gives Person A Chocolate Bar E

Thing A verb Thing B

 ——→

Person A leaves Shop B

Thing A verb Thing E

 ——→

Person A eats Chocolate Bar E

In order to synthesise these occurrences to see if there is a system we need to start to pull all these individual events into one time independent model

In the example above, we have already given the tag of person to two people, so there must be a class called person. We have been told there is a class called shop. But we may, by interpreting what happens in the shop see that there is a role being played here by people within the shop – and we learn of a new class – a customer and a shop assistant. We may even notice that the same person can be a customer in one shop, but a shop assistant in another.

I have tried to show this diagrammatically below.

I have added just a few arrows to show how we would put the information in the boxes.

What have we deduced about the system? A person can be a customer or a shop assistant in a shop. A. person may be a customer in many shops and a shop may have many customers. It may also have many shop assistants. A shop may have a number of products in the shop and in the shop the product may have a certain cost. Conversely a product, such as a chocolate bar may be for sale in any number of shops.

This is our CLASS and RELATIONSHIP model it provides us with the placeholder for the occurrences of things.

Person and A and Person C are both People, but Person C might also be classified as a shop assistant and person A as a customer in that context. Chocolate Bar E is a product in that shop which has a cost dependent on the shop

Next we find that the chocolate bar is but one of many types of chocolate bar and that along with chocolate bars, in the context of shopping, there are other products which can be bought – books, magazines, cigarettes and so on. All these are found in shops and can be bought for money. Then we notice that the amount of money paid in each shop for the same thing can be different and so we start to associate something called 'cost' with a product in a specific shop.

We build up a picture in our mind of classes of things and how they inter relate and also what sorts of things we can know about them. What do we know about books – for example? Well they have pages and text, a title, an author or authors and perhaps have an ISBN number and so on. When we obtain information about a book, we may store this information in these ‘place holders’. In this way we make knowledge more accessible.

What we can thus see is the process of synthesis in action. We do build on existing systems, adding extra classes or subclasses [book, newspaper, cigarette, confectionary, chocolate bar etc] to the existing model and adding new relationships [has, is, becomes,] as well as attributes [number of pages in book, ISBN number, title]. And we gradually populate the model with more facts that help to reinforce the model.

I think it is also clear why we prefer to build on an existing model rather than build up a new model from scratch, it is far quicker, far easier and also emotionally more satisfying to build and reinforce rather than reject and start again.

Children don’t mind throwing away and starting again because their little edifice of models is still very small, but the older we get, I’m afraid the more reluctant we tend to be to start again…

Below I have shown the example model again.

What we should be able to see is that the place holders act like indexes to knowledge. Thus if someone says 'shop' we may then think of all the shops we know about because we have been there as a customer and indexed the images and information about shops within this place holder. Similarly the idea of the shop may lead us to think about the products the shops sell. I have shown this diagrammatically below. This incidentally is exactly what happens in databases in computing. There are great nets of information all linked up by logical links.

EXAMPLE

 

The more experiences we have, the better our Memory is and the better we are at classifying we are – the closer to 'reality' we will get, but we will never actually be able to build up a complete picture of reality. 'Reality' – the complete and accurate model - only exists in the external spiritual world.

In the same way, our model will not only be different, but our 'tagged occurrences' will also be different. In effect, we may or may not remember the names of things and may or may not be able to make the links between them. The diagram below shows the occurrences. We may never have seen that Tesco stock Hartleys Strawberry jam and Nescafe Gold Blend – as do Waitrose and probably LIDL – because we have never consciously recorded this information.

What you are seeing above, by the way, is the classic database layout and diagram. Databases in computing hold the information like this and make all the links for you, so this is why databases are so good at 'remembering' and then making associations between data. It is why you can ask questions like 'tell me if Jim Jones has ever been a customer of Waitrose' or 'tell me the price of Cadbury's Dairy Milk in LIDL'.

The Databases of functions 

In the diagram we have simply shown how we have classified some of the data or information. But this is only one half of a system, a system also has action, function, activity so the classification process will also be a process of trying to deduce if there is a system to the activities as well.

In the example, the activities we see over and over again of those of 'buying' and 'selling', then we see there are activities related to running a shop such as costing products, obtaining products from manufacturers, stocking shops, and so on and we learn how shops are run.

 We are learning about system and deducing what system is in operation from the observations we have made and everything hinges on how well we have tagged the original occurrences – the untagged observations.

Without going into all the detail I went into for the database of facts, I think we can also see that we should be able to deduce a system exists for buying products from a shop, in simplistic terms.

This is an activity dependency diagram. If we had been trying to describe activities with data dependencies we would have been able to have drawn a data flow diagram – and a data flow diagram is what I have been using all along to describe these processes of the mind – memory, learning, perception, will and command and control and so on.

All systems can be described using these sorts of diagram and although I am sure we don’t hold diagrams like these in our minds, I am certain we hold the equivalent information if we have learnt a system properly.

So from the basic observations made – whether we read about things or actually saw them we end up with knowledge about systems and facts about things. Tagged classified and sorted into hopefully accessible placeholders.

This is learning and what learning is all about. The end result is what we call 'knowledge'.

Knowledge is stored in our personal memory. Thus memory is derived knowledge of systems and data, whereas perceptions and thoughts are the raw observations from which memory is obtained.

The database of images 

Up until now I have concentrated on showing how synthesis is used to build up systems information – databases of classified facts and a database of learnt function. But I think it is also important to mention that we also build up a database of images. These images are used in recognition.

We build up in our memory a whole host of 'templates' we have constructed using our 5 senses of what we mean by a 'something' . An image, in effect, a blueprint or an archetype. We construct a picture of 'reality', which is the sum total of the input from our 5 senses and then classify it. In effect, we don't just build up belief systems in memory, but we also build up archetypal models that help us navigate and make decisions.

Memory is thus used to build a picture of our 'reality', not just to create a sense of system.

Notes

For some additional detail about what can go wrong during synthesis see Learning - synthesis, what can go wrong  

Observations

For iPad/iPhone users: tap letter twice to get list of items.