13th Jan. 1985 Ideas from reading "Dynamic Memory" by Schank

Instead of storing feelings whenever attention is paid to them - maybe they are stored in parallel with those items attention is paid to. Therefore attention is not paid to feelings - it is more of a status indicator always associated with all inputs.

Maybe goals - wants are kept in short term memory so they are semiconscious during processing - rather like a status indicator as above - maybe feelings as above are kept during short term memory life time?

By combining - by executing proved sequences in order a new memory with a longer proved sequence is produced.  Certain frames must be superfluous because they are not relevant to the re-execution. Thus they would be skipped over at association time and execution time. Re-execution could be set off in parallel for a proven sequence.

Is ignoring commonly heard sounds - i.e. forgetting what was being played on the radio when you drive to work analogous to automatic output except on the input side. - Probably not because?  Yes, because attention is not involved - it takes place in parallel - If it is unusual it interrupts attention.

Goals are inputs. We have situation + action -> outcome. The outcome is the goal. The situation is an input and an action is an output. The outcome is a situation also. ==>> Input + Output -> Input. Memory has many such sequences.

This implies that rewards are associated with inputs. The final input is the goal and it is either neutral, a good goal or a bad goal!

When below concentration level of directed thought and action then any sequence that has been seen or done before causes lack of interest and a desire to change the world, which means reflexive output.

When during concentration level of directed thought and action any sequence of past experiences is violated then attention is directed here because it is not expected sequence.

10th Feb. 1985

The following is a list of the expected improvements to the simulation from incorporating recognition of a series of "ideas" rather than just the one "frame".

  - Recognize context and act differently on same cue depending on the context.

  - Recognize a series of inputs - act upon recognition of whole series not just single input.

  - Keeps all series that match 'active' so easier to identify which is same - easier to identify if it’s a new situation and therefore needs attention.

  - Emulates hierarchy of inputs as mentioned 21st March 1981.

  - Helps produce generalization and over discrimination as in 21st May 1981.

18th Feb. 1985

Just as "an input is only stored when attention is attracted and attention is only attracted by a change in the environment" a sequence of inputs should be ignored until something different - a change - occurs. This means attention is attracted to a change in the expected. - Likewise only the change is recorded. Does this also apply to emotions, only recorded when a change in emotional state takes place?

14th June 1985

I am thinking about automatic output and wondering how can it continue through a sequence of outputs - It needs a continuous supply of cues. These must come from a sense - touch + muscle receptors. [ a comment added 21st May 1986 – These handled by cerebellum in man! ] These are senses which we are not conscious of while the automatic output is happening because we are concentrating on other senses - sights, sounds etc.

 So when we are concentrating - learning some sequence of actions the sequence is cued on sights and/or sounds - and the automatic output takes its cues from feeling (physical) which also happen while learning.

 When doing something automatically i.e. walking while doing something else which requires concentration i.e. reading the automatic process can not continue if it needs the sight for something such as stepping off the curb.

15th June 1985

A sequence of Input -> Output once learned does not become habitual if there is no associated reward or idea of reward each time it is done. So if their is an Input -> Output (un-rewarded) sequence that is known it is not performed because it does not produce any new inputs i.e. curiosity is not satisfied (see bottom 16th Aug 1981).

From 25th Dec 1979 - When we have a dynamic environment the input changes - however boredom can still result if there is repetition in the changing environment. A changing environment like a play, movie or TV does not repeat and therefore is not boring - so one can stay absolutely motionless and speechless while being "entertained" - except repeats of commercials.

You are faced with a situation and you think through a sequence of input and outputs until you find a goal you want to achieve - now how do you put the "thought plan" into action. Well the current situation is still facing you and this is the cue for doing the sequence you just thought of - which is still in short term memory.

On 18th Feb. 1985 we say input only stored on attraction of attention but it is also stored on purposeful behaviour when we purposely look at something - execute a sequence at a high level of concentration. Often we are looking for an expected thing to happen.

Maybe we find it difficult to read and walk at the same time because walking and reading both require the automatic output processor - one to walk - one to scan lines.

Must change in the model:

 - Reward / Punishment only associated with Inputs (not outputs).

 - Idea of Reward & punishment only associated with Idea of Input and Idea of Output.

 - Recognition of a sequence of inputs.

 - Execution of proved sequences - skipping non relevant frames only paying attention to necessary frames to do sequence execution not recorded - only initial cue.

 - Recognition of repetition of any length sequence and resulting boredom - need to do reflexive output.

 - Recognition of repetitive idea sequence and boredom -> giving reflexive action.

 Undated Notes

Working at one level can be interrupted by same level initiators.

Automatic execution - can only be halted by pre-set interruption - condition which attracts attention.

(SKIP) Recognition of sequence of senses not interrupted over time i.e. recall of sequence of inputs.

(DONE) Reward and punishment only associate with input- output sequence.

(DONE) Idea of reward & punishment only associate with idea of input and idea of output sequence.

(Done) Punish reward - new attention state.

(Done) Reflex output when punishment - avoidance, move i.e. reflexive output (9th June 1977)

(Done) Blank when no recall found - interest - drive, see next

Curiosity as a drive. Whenever two input or attentions + data are identical output something, or input completely new -> interest ---?--- generate output - outputs last input (Maybe).

Imitation reflex when executing output but there is no associated output pick up input and output it.

(No) Punishment comes from N - generated internally.

(Done) Reward is associated with Y. - internally produced.

Must define before hand the environment adaptron sees whenever it outputs something then return input is defined.

Add thinking

Sound made comes back as input. - What level does it interrupt? Maybe heard only if next attention is paid to hearing.

If doing output and no associated output then take associated idea of output. If none then take Associated input else blank?

When you start to execute because of associated reward recall results from storing reward do not attract attention, do not wipe out other recalls.

In most ways reward is always reflexively stored and never has any associations. When executed no recall is done. Blank returned if no value.

3 types of recall needed - remember it and get the associations, record it and clear only reward, record it and clear all - no carry through of previous associations.