Suite 302 - 40 Boteler St.
Ottawa, Ontario, Canada K1N 9C8
E-Mail: Mail(at)AdaptronInc(dot)com
Phone: 1-613-218-5588
 
Adaptron Inc.     Research   Notes   Papers
 
This page updated: Feb. 27th, 2013
Research - Notes: 1993
 
3rd March 1993
 
Idea from reading 21st May 1986 is that there is a stream of consciousness which simply processes experiences in sequence unless interrupted. So at no concentration level it recalls sequentially, like singing a song. The sequential mode plays back experiences in order as a default mode of 'thinking' when nothing is happening to interrupt it. And we have also learned to do this when we need that sequential play back capability.
 
15th March 1993
 
2 types of doing that might run simultaneously:
- Action doing
- Thought doing
 
Next to match sequence in pattern recognition follows same rules as association links.
 
When a thought sequence repeats and we are not doing or paying attention then this equates to boring and some action is taken.
 
Interpret model (experience) and decide what to do either same or different. If different and no association available then generate output.
 
There are two types of things that the mind can be doing: action things which result in some detectable activity and thought things which are the processing of thoughts. They can be going on in parallel. The activity only gets interrupted when the thought process results in the decision to do some activity that requires the busy sensors or output devices.
 
As we are interpreting the experiences in memory (That's what thinking is - interpreting a model of reality which is the collection of all the experiences of the being) which we do by playing back the experience and looking for the emotional associations we make a decision that we either want the same as what our experience tells us or we want something different. If it is something different we need to perform some other activity than was done in our experience. This is where we have a feeling for 'what can we do?' and if something is recalled then that is done else we need to generate some random output.
 
When a thought sequence repeats and we are not concentrating / paying attention or doing a previous activity then we have a "boredom" event and some activity is initiated.
 
The strength of a recollection is reinforced by events happening in the same sequence as stored in memory. The sequence matches, continues to match the memory trace provided the inputs match up with the trace where the paths, links through the trace are possible. Links that can be used for pattern matching the sequence are the same ones; follow the same rules that are used to determine if one thought / recollection is associated with another.
 
4th April 1993
Re: 15th June 1985, we probably get that squirmy feeling when reading and walking at the same time because both need vision to work and walking needs it so we don't end up flat on our face (punishment). However if you were walking on a known deserted road you can tell yourself to do the 'walk without paying attention (vision)' operation and then do it as a repetitive leg action not needing vision while dedicating your vision to the reading.
 
A drive is persistent in time until it is satisfied at which point it goes away. It later on reoccurs and again persists until it is satisfied. Are mental goals such as the desire to reach a particular outcome which you know you will recognize when it occurs similarly persistent in time? Can you have only one such mental goal at a time and therefore one is interrupted by another? Or is it that at any one time one goal is current - in awareness and the other ones are recalled and thought about - especially when trying to achieve a sub-goal of a more global desired goal?
 
Given that the experience is a limited knowledge base of the real world upon which to model the world for thinking about what to do we need a recalled feeling from memory when a purposeful, concentrated thought operation is being performed looking for an association and none is returned. This then is the feeling 'I don't know'.
 
5th April 1993
 
Fear is a strong emotion caused by the encountering of a situation that causes instinctive fear or the expectation of a negative thing about to happen. This fear is a primary input from the pleasure / pain centers of the brain when the situation is recognized. There is also the idea of fear (recollection of fear or fear recalled) which helps guide our thinking and consequently our actions.
 
11th April 1993
 
The brain must have some internal states which it can recognize as concepts and which it can use to control its thinking processes. For example in directed thinking when one is trying to come up with an associated idea or when one is asked if one knows something the memory/recall process returns a concept of not found/ don't know.
 
The phenomena of 8th June 1972 - 165 (this section) could be explained by the minds ability to directedly execute associated idea attentions but not directedly execute next frame attentions.
 
24th April 1993
 
Emotions are the external manifestations of feelings.
 
Primary feelings are produced and attention is paid to them by a pleasure / pain center which is always alert to certain patterns of inputs. The idea of feelings is then used to guide actions and thinking.
 
25th April 1993
 
I am only interested in the explicit, declarative types of learning which require a conscious record/memory to be made. (As opposed to the implicit, non-declarative, subconscious/nerve level learning that can take place)
 
Undated - Association processing replaces short-term memory processes.
 
8th May 1993
 
Another internal concept might be 'latest accomplishment does not satisfy outstanding goal desired - unsatisfied - expectations not met'. This goal matching recognition is mentioned in 15th July 1992.
 
11th May 1993
 
While attending the local SIGART meeting I thought about how one could deduce that an entity was able to learn and / or think just by observing that entity, i.e. via its behaviour.
 
 - The environment could be classified into a number of types:
 1/  linear in time - sound (pitch & volume) or light (colour & brightness) or letters / words typed.
 2/  2 dimensional - sight / projection in time
 3/  2 dimensional physical - motion / obstacles (touch)
 4/  1 dimensional physical - motion
or   A/ Passive - environment does not change unless active agent / robot makes change.
       B/ Active - Things in environment which change location etc.
 
 - Motivations are important to be understood if some meaning is to be obtained from behaviours. There are positive and negative ones often the reverse of each other. Positive: Food (Energy), Interest - mental challenge - sight/sound, Pleasures - biological - aligned with reproduction and energy.
Negative: Avoidance - heat/cold - survival, no air, pressure - body damage, hunger - pain (low energy), Boredom.
 
 - Observe learning taking place when it avoids painful situations! - Survival, repeats pleasurable activities.
 
 - Observe thinking taking place when it explores the environment - Interest, Does not repeat exploring known environment - or exercises/uses known environment to get somewhere (not to explore it), seeks out pleasure.
 
16th May 1993
 
Punishment/reward not associated with response - only situations are good/bad - responses are never bad in themselves, the context in which they are done determines whether a bad stimulus occurs.
 
Avoidance of boredom is pursuit of interest/change/differences. Does this mean that in the same situation as before does a robot experiment with new responses? Maybe it does if the old ones are not rewarded and not punished i.e. no feelings or idea of feelings obtained in the past situation. As long as it got an idea of reward then the sub-goal is being achieved in the past because the idea of reward feeling is obtained when the goal is achieved. For goal seeking to exist it needs at least one thing to have been rewarded. It needs this before it can come up with goals and sub-goals. Without this reward it is just trying to avoid boredom. Ideas of reward chain backwards over time through repetitive associations so that long chains of sub-goaled executions exist all focusing on at their end the rewarded state. S1 - R - S2 (Rewarded idea) when recalled on input of S1 results in S0 - R - S1 (rewarded idea chain sub-goal backwards one step.)
 
The normal stimulus response episode is S1 - R1 - S2 - (Reward Idea). This means that when S1 reoccurs and associations are being obtained at recall time R1 and S2 are strongly associated and an idea of reward is associated too. Attention is first paid to the idea of reward that is stored after new S1 and execution begins at the R1. This conflicts with the idea of recalling R1 or S2 and starting the thinking process of recalling ideas and storing ideas.
 
The strength of an association is determined by the reward, idea of reward, neither or punishment idea of punishment that follows the associated item. This is used at attention attraction time to identify which association should be attended to first. But this strength does not in any way provide the signal that the recalled sequence should be executed. It will be the associated feelings that will initiate execution, which will most likely be there anyway in order to get the strength of association.
 
Undated - Need an attention function that is 'next' as well as other associations.
 
 - Input associated with output only in the original input context.
 
29th May 1993
 
With reference to 29th Nov. 1972 CAT has many associated concepts whereas XCQ does not.
 
With reference to 13th May 1970 - 135 - the immediate memory is cerebellum memory, knowledge of sequence of dance steps that can be repeated without conscious thinking. Thinking can go on in parallel with it to cue the dance sequences, give it changes in direction.
 
14th Sept 1993
 
Typical sequences of learning events resulting from rewarded behaviour propagation are:
 
  S    S      R       S        F
bell food eat  taste good
 
S= Stimulus, R=Response, F=Feeling, Fi=Feeling idea. Stimulus invokes the feeling, not the response.
 
  S    S      Fi      R     S        F
bell food good eat taste good
                   ^                         ^
               execute              halt
 
  S    Fi       S     R     S      F
bell good food eat taste good
           ^                                 ^
       execute                       halt
 
Punished behaviour - propagation cancels learned behaviour:
 
  S      Fi      S    R      S     F
bell good food eat taste bad
           ^                               ^
       execute                     halt
 
  S      Fi      S     Fi   S
bell good food bad   ?
           ^               ^
       execute     halt
 
  S      S     S
bell   food  ?
 
If there is punishment only:
 
  S    S      R     S       F
bell food eat taste bad
 
  S    S     Fi     S
bell food bad  ?
 
This scenario assumes agent in control of responses and stimulus is a result of the responses. There are some environments where you loose control - doing nothing results in punishment. How does this stimulate the agent into doing something - avoidance etc.?
 
  S    S      R     S      F     R     S       F
bell food eat taste bad spit taste good
           ^                     ^
     controlled        reflex
     response        reaction
 
or consider:
 
  S      S                S         F      R           S             F
bell red-zone electricity bad leap green-zone good
 
  S      S            Fi    R        S         F
bell red-zone bad leap normal good
 
In these sequences we do not recall Si and then Fi. When do we produce Si recollections?
 
8th Nov 1993
 
Some dictionary definitions help to get our mind around what is thinking.
 
Intelligence: ability to perceive logical relationships and use one's knowledge to solve problems and respond appropriately to novel situations - Not acquired knowledge - perform human reasoning.
Think: Process of arranging ideas in a pattern of relationships or of adding new ideas soon to be related to such a pattern - imagine.
 
My definition includes the processes of reasoning, sound-intelligent reasoning, problem solving, deduction, induction, abduction? not observable = covert (not overt), requires internal symbolic representation, requires an inference approach or sanctioned and recommended set of inferences based on the symbolic representation.
Reason:        Logically think - to understand - to draw inferences.
Induction:     Drawing a general conclusion from number of known facts.
Deduction:   Reasoning from the general to the particular. Infer.
Inferences:  Arrive at by thinking - to deduce, forming conclusions from premises.
Ideas:           Mental images, conception, opinion, plan, notion conceived by the mind
Concepts:    Thought or opinion, general notion or idea, especially one formed by generalization from particular examples.
Learning:      Get knowledge of (subject) or skill (art etc.) by study, experience (and practice), or being taught (instruction) by observation. Commit to memory.
Notes: 1992   Notes: 1993 - top   Notes: 1994
 Adaptron Inc.   Research   Notes   Papers