Any tricks for short-term memory during calculation?

I believe one critical part of visual calculation is remembering where you left a unit that you just mentally moved. I have some trouble with this part of calculation, and I suspect others do, too. Psychologists call this type of memory "short term memory" (STM) (http://www.human-memory.net/types_short.html), or sometimes "scratchpad memory" after the computer science term (https://en.wikipedia.org/wiki/Scratchpad_memory).

Through experimentation I've found that the problem gets worse in the following situations:

(1) on an open board containing few units, because all those squares tend to look the same (no distinguishing "landmarks" like pawns near them)

(2) on an open board where the units being moved are infinite-range units (queens, rooks, and bishops), because remembering the general region where they're moving is of no help (unlike the regions of kings or pawns)

(3) where there exist many units on the board, especially when those units are interacting, especially with newly opened lines of attack, because there is so much to remember

Lately I've been experimenting with (1) and (2), and I've found a trick that helps me: I remember the color of the square to which the unit has moved, and I mentally call out "white" or "black" as I move the unit there. That reduces the number of squares to consider at when I look back to the region after moving the other side's unit, to remember the square where I left my king, for example. I've found that trick to be particularly useful in endgames with pawn races and/or king races.

This trick of mine works surprisingly well with kings--case (1) above--to the extent that I can basically do infinite lookahead that way, which is very encouraging, but as soon as I switch to mentally moving rooks, I start to have the above-mentioned problem again: because the possible landing squares are so numerous, mentally remembering "white" or "black" often leaves me unsure as to the rank of file where I left the unit.

So my questions to others are:

Do you have a similar problem remembering where you mentally moved a unit during calculation, especially in case (2)?

Do you have a trick that works better than remembering the square color?

One obvious solution is to remember the exact square name (e.g. "b4") instead of its color (e.g. "black"). I haven't tried this, but I suspect this method will be problematic because (unless one practices extensively) it requires an extreme familiarity with square names to avoid having to stall for a few seconds to determine the square name with each ply. However, maybe such pre-practice for square name familiarity is worthwhile in the long term. 

There exist intermediate solutions, too, like dividing the board into regions, or remembering in which direction the unit moved, or maybe just one of the rank/file designations (maybe just the new rank/file designation?), but I haven't experimented with these methods yet. Another interesting phenomenon I've noticed is that it is often easier to move units quickly rather than slowly, because short-term memory hasn't had enough time to fade out, but I haven't experimented with this, either. I'm hoping somebody else has a memory trick or two that is known to work, so as to spare me extensive experimentation time.

I don't know any great answers other than practice.  As for remembering square names, it really doesn't take long to do so.  Just play a lot of games taking annotation and study old masters games by reading the annotation.  After a short while, the squares should become like old friends ;)

Exactly. I've heard of such online training before, and my immediate impression was: "How stupid! What does learning square names have to do with playing better chess?!" There's no doubt I could learn the square names well, but what I'm wondering if this is the best approach. I wonder what masters do, for example. Even if that's what masters do, maybe there's a more efficient way to do the same thing. Learning all those square names very well will take a long time, and I wonder if it's worth it.

Somebody on this forum once mentioned that Bobby Fischer's original score sheets were full of errors, and that's what mine are like, too, especially when playing Black when suddenly I have to invert everything: having to invert is just one more step and annoyance along the way to what seems to me should be an easy task. My point is that (if that story is true) Bobby Fischer evidently didn't know the square names much better than I, but his chess skill was still superb.

After a few moves I visualize the area or whole board, as if it were a new position.

I think a good way to train this is set up a board, calculate just a few moves (lets say 2 or 3) then try to visualize the new position. Notice any new open files, new pawn structures, mentally locate each new piece, etc.

Then when you have it about as good as you can get it in your mind's eye, play the moves and now you get to actually see the new position. What relationships did you miss? For example now your queen is attacking an undefended pawn, or now your opponent can play a check, things like that.

This is also a good way to remember where a piece is, if you're having trouble. Give it a job. "It's attacking ____ pawn or defending _____ square" etc.

They do have that here at chess.com. It's called "Vision" under the "Learn" link in the menu bar. It's actually really great and has several options that involves only the squares and/or pieces.

My best result is 22 out of 23 squares.

https://www.chess.com/vision

Your first suggestion is roughly what the book "Chess Training Pocket Book II" (Lev Alburt & Al Lawrence) recommends: keeping certain key positions extra heavily in mind during visualization. He likened it to creating platforms at key positions on the tower you're climbing, so that if you lose track you can fall back to one of those platforms lower down on the tower.

Your idea of "It's attacking ____ pawn or defending _____ square" type of relationships might be a great idea: I'll have to think about that. Thanks. That makes more logical sense than mechanically remembering square names since such abstract features (lines of attack, distance that a rook's attacking file is from the passed pawn, etc.) are the important features since they enter into the overall plan already, more so than remembering the name of a rook's temporary stopping point.

I think it helps to go slow... sometimes I remind myself that before a tournament game. It's really easy (at least for me) to quickly calculate 4, 6, 8 or more moves without actually visualizing. The end position is fuzzy, I don't quite know where things are, but the last 1 or two pieces that moved I know where they are and whether they're safe from being immediately captured... hope what I'm saying sort of makes sense, not sure if I'm saying it well.

But to go slowly means you're actually seeing the new position every few moves. You could, for example, survey the imagined position to identify all the undefended pieces, or all the moves your opponent has to give check. This is a lot harder than calculation without visualization of course, and depending on the position sometimes I can get easily confused only a few moves deep.

Visualisations can be very quick when they come "automaticly", like sequences of different solutions that flash into one's mind when looking at the board. I do, however, have to analyze my own visualisations to see if I have missed something, like missed a possible move from the opponent, or missed the actual position of one of the pawns.

I understand what you're saying, I believe: the end position is fuzzy except for the most recently moved pieces. The other pieces have already begun to fade from memory. But the fuzzy position could be generated quickly.

Maybe I'll use this opportunity to describe an experiment I've been trying... Since I had such good luck with my mental announcements of "white" or "black" when moving only kings, I conjectured that an underlying problem was that having only two colors made it too easy to get the position wrong if I wasn't sure of the region. After all, two colors span only two squares before repeating; two colors have a resolution distance of two. So I mentally overlaid a translucent yellow color over 4x4 regions across the board, in a checkerboard pattern, over a blue-and-white board. This turned half of the dark squares (blue) into green, and turned half of the light squares (white) into yellow. This system made the new regions twice as large before repeating; four colors have a resolution distance of 4. Since four is half the board, it seems it would be hard to forget which region it was.

I then practiced visualization using this 4-colored board on endgame positions where the board was mostly empty. It took two days of my spare time at work (I look at chess puzzles when I'm not busy at work!) before I got so familiar with this color layout that I could immediately visualize the color (out of the four colors) of any arbitrary square. I lost interest since then, but it seemed to me that system was working quite well while I was using it. In that system I would then mentally call out "blue," "white," "green", or "yellow" as I mentally set a unit on a square.

Since I haven't experimented with this system very long, I can't guarantee it's a good system, but maybe somebody here might want to try it and report back. Here's a quick diagram of the board coloration I mean (I mocked this up in Paint just now, so it's only a rough diagram):

I don't understand. What exactly are you doing here? Are you randomly moving a king around the board and then stating whether the square it's on is white or black?

If so, what is reason?

I look at the White king, decide where I want to move it, move it there (mentally), call out the square color, then look at the Black king, decide where I want to move it, move it there (mentally), call out the square color, then I repeat the process, indefinitely. It's during that switching back-and-forth process where I sometimes lose track of where I placed a rook, queen, or bishop, *if* any of those other pieces were involved. I don't have any trouble remembering where the kings were, presumably because the combination of approximate region + square color narrows down the exact location quickly and sufficiently.

Why am I doing this? I'm experimenting with human memory during chess and using myself as a guinea pig. Why? Because the resulting insights are important and fundamental, not only to chess but to psychology, and because nobody has done such experimentation before, at least not to my knowledge.

I found out that memorizing square colors or playing blindfold chess does not help much one to increas skills - infact all that calculation and visualization drains so much of your energy and you might blunder horrible in long time control game. it happened to me. I was playing higher rated opponent who could do blindfold chess and calculate very deep. 

We go to fide 60+30 tournament and pairings matched us, he blundered twice and lost the game whihle most of the time i was sitting watching him concentrate so much and calculate soo much on obvious moves.. he was in time trouble and angry cuz he though he was winning easy against me.

I would say that just practice a lot, and it will probably get better. After you do a few weeks of intense visualization practice, you'll see if you really need any tricks or not. 

Thanks for the experience-based insight, AllviewP9Lite, and for the suggestion, Cherub.

An update: Recently I've become more aware of the importance of short term memory of shapes/configurations for visualization. I noticed this when looking through several of Pandolfini's endgame positions one night at work: although I didn't have trouble mentally moving the pieces, I did have trouble remembering the resulting configuration of units, like was it a rectangle? key shape? cross shape? L-shape? etc. I noticed have a tendency to overlook uncovered lines / discovered attacks as a result of that weakness. That leads me to a hypothesis that possibly a good way to practice visualization is not so much mentally moving pieces, but rather quickly remembering the shape of a group of pieces so that the shape can be recalled accurately when the picture is removed. This is particularly noticeable in pawn endgames when there exist few units and they tend to cluster around the square where the pawn will promote. This hypothesis could be tested easily using a video or computer program, though as usual I absolutely don't have time to work on it.

I still say that sheer practice is a cop-out. It works, but we don't know why. I want to know why. Until we know that, we aren't in control of our learning, and therefore we cannot make visualization learning any more efficient than it has been for the last several hundred years.

 Source: Pandolfini, Bruce. 1988. Pandolfini's Endgame Course. New York, New York: Simon & Schuster. Page 61.