Commenting to comment

 Robert commented on the last post. I will comment to that in blue.

Such considerations suggest an important difficulty our cognitive processes are able to address: the total information contained in an environment is too vast to process using deliberative reasoning but the information contained in localised chunks of the environment is too focused to be useful by itself. What we would like to find are the small number of visual cues that make up the salient aspects of the environment and show how these cues change as a function of skill. One approach is to use a neural network that can automatically form ordered, compressed representations of sensory perceptions such as Self-Organizing Maps (SoMs). SoMs have been used as a model of neurological organisation as well as a tool for data-mining and they have the benefit of being unsupervised learners, that is to say they extract structural regularity from data without external guidance. This last point is significant, from a behavioural perspective human players implicitly learning relationships between game pieces are not aware of what is being learned, they are only picking up on the statistical regularities [without any awareness of statistical properties] in the task environment, this requires an unsupervised process.

How do we find the “Goldilocks zone” between ‘too vast to process’ (looking at the entire position plus the tree explosion while looking ahead several moves) and ‘too focused’ (mis-perceiving some salient aspect(s) of the position as the crucial aspect to focus on)?

We should not worry about too vast to process too much. There is a difference between the study room and the tournament hall. In the study room, nothing is too complicated, since we use Mr. Stockfish as our tutor. You are never going to see exact this position in the tournament hall, so you only have to worry about what you need to learn from this position that is reusable in the tournament hall.  

In the given example position, what are the salient CUES?

I assume it is our addiction to system 2 that makes us ask such questions. We don't know what the needs of system 1 are, and how visual cues are build. That magic lies in the realm of system 1. 

I suggest the following process that readily comes to mind just looking at the position as-is:

1. (1) The white Queen seems poised to go to h5, simultaneously threatening mate on h7. After “seeing” this possibility, it MAY trigger the additional perception that f7 would then be B.A.D. (whereas before that move, it was “overprotected” by black). This also takes care of the ‘concern’ about the attack on the white Bishop; a mate threat is an Equal or Greater Threat that forces black’s reply.
   
   (2) Kicking the black Knight off of c5 seems somewhat less salient at first glance. That might trigger the perception that the black Bishop on b7 would then be unprotected. It did not trigger my thoughts in that direction and certainly NOT connected to the concluding Rook fork.
   
   (3) Perhaps it is here that the ‘idea’ occurs of lots of capture occurring on f7, leaving the black King sitting on f7, and vulnerable to a check by the white Rook.
   
   (4) This may trigger the idea of dropping the white Rook on d7, forking the black King and Bishop.
   
   (5) The idea of a potential fork on d7 might have brought up (again) the idea of kicking the black Knight off c5, leaving that Bishop unprotected.
   
   This somewhat mixed (intuition plus logical thoughts) process seems similar to GM Beim’s idea of the examination [and assessment] of the position (the vulture’s eye view), supplemented by short-term tactics (1-3 moves, not more). That process is not a strict calculation of variations, creating a Tree of Analysis a la Kotov, religiously examining each branch of the tree without repetition. 

   We must distinguish between system 2 in the study room and in the tournament hall. Your five point list above is typical for system 2 in the study room. Your only worry for now is to educate system 1 in the study room. Since that is what you are going to take with you. All system 2 considerations remain in the study room. There is a special task for system 2 in the tournament hall. That is totally different from its task in the study room.
2. In the study room, we seek to grasp the position as deep and good as we can with system 2. Our attention is the spotlight which enlightens system 2. System 1 looks over our shoulder and works its magic. Since system 1 can only work by the light of attention.
3. In the tournament hall, there is a totally different situation. Here we are totally depended of what system 1 comes up with when we look at the position. The role of system 2 is completely different here. System 2 has to check whether what system 1 comes up with is correct. It makes use of the tree of scenarios to do so.

[Herbert] Simon summarised his results in the following way: “The situation has provided a cue; this cue has given the expert access to information stored in memory, and the information provides the answer […]” whereby “[w]e are aware of the fact of recognition, which gives us access to our knowledge […]; we are not aware of the processes that accomplish the recognition.” [original emphasis]. The goal of this work, then, is to find the perceptual templates that amateurs and professionals have acquired through perceptual learning and that they implement as the basis of their perceptual expertise when playing . . .

For training purposes, we should factor out the simpler ‘patterns’, and then focus our attention on the relationships between those patterns. It is the recognition of the component element patterns AND the interrelationships between them that needs to be stored into System 1 by training.

Maybe it is more simple than that. What do you need to see? 

• A series of forced exchanges where the king is ending up on the exchange square. That is a very common theme. You only have to recognize it
• Then you have to recognize the potential rook fork
• Then there is some little work to do for system 2: why is the fork not working out and can I make it work?
• The tree of scenarios tells us that the knight defends both the bishop and the invasion square. And that elimination of the defender is the natural next step 

As long as you do not see the solution as simple, system 1 has not fully absorbed it. You have to see the full picture. Your point of view has to change. The sage who looks down from a mountain to a train that drives by, sees something totally different than someone who is in the train and who is looking out the window.

Is there a guarantee that going through the examination/assessment process using short-term tactics to connect the elements via interrelationships will put these factors into System 1 in such a way as to trigger pattern recognition in similar (yet unknown) positions in the future? NO.

The point of view from a distance is a guarantee that you see the interrelationships well.

However, given a sufficient number of exercises that are similar in some salient way to the elements/interrelationships identified in THIS position is almost guaranteed to eventually create a generalized abstraction that WILL be recognized. That is how skill is developed over time – focused attention on the cues that are necessary to “see” the contours of the essence (spirit; demands) of the position, and the method(s) to be used to combine those elements into a coherent combination.

Unfortunately, we have no empirical data regarding how many exercises (or the combination of appropriate elements) which will be required in order to subconsciously create the cues that enable us to “see” almost instantly what the salient characteristics (chunks/templates/schemata) that are salient in an unfamiliar position. There are vertical (higher levels of abstraction) as well as horizontal (sufficient numbers of similar examples) dimensions to the training required to build skill.

We have empirical data, though. I use a database of 722 well chosen problems. And after 240 hours of training, I recognized at least two positions in 16 games.

MDLM used ct-art, which has around 1100 positions and gained 400 points in 400 days.

Judith Polgar became grandmaster when she was 12, so she needed about 8 years for that.

There are 32 mates, 6 tactical themes and 7 types of preliminary moves. 

Memorization of this specific position (pieces-on-squares) is highly unlikely to create a recognition template in LTM. The requirement is repetition of the process of “seeing” (identifying and focusing on) the basic elements/cues, not solving large numbers of problems at warp(ed) speed.

That learning process is not nearly as much fun as playing Puzzle Storm!

I'm not familiar with Puzzle Storm, so I don't know about that. If you can flick in some serious study in your puzzle storm AND the elements in puzzle storm have a high frequency of occurrence, you might transform it into something useful. Deepening my knowledge of the ABC has become audible by the many aaahs and wows in the study room lately.