@Carlo1310 said in #27:
> If I understand correctly, reults in shape and scrambled conditions do not correlate with rating, while standard condition does.
>
> An anectodal observation of qualitative nature: I've always found it to be striking, how higher level and expert level chess players tend to guess and talk about positions which opening they arised from or "look like". Like, "this looks like it was a semi-slav" or "this may have arisen from a benoni structure", etc.
>
> One possible factor that helps them complete this task is not only the resemblance of these 'natural' positions but also the type of different memorization resources they can pull from to re-construct the position: Have they experienced the position, what did they classify it as, helping them to more accurately assess the overall pawn structures. Or helping them to memorize the traditional (read: often-seen and -analysed) aspects of a position as one or a few chunks, while making a more conscious effort to also memorize the specific "exeptions" in the position presented, such as a displaced queen, an unusual pawn advance, and so on. So basically, as if each position was possible to be deconstructed into a few chunks + exceptions and some important positions may just be a prototype chunk in itself altogether.
>
> Circling back to the observation about the guessing game of structural recognition stronger players tend to "play" casually, there would be no benefit that I can identify on the surface level, other than showing off (unlikely, when you face these types of persons and statements so often, even from some of the most unpretentious persons I know) if it wasn't for an advantage in playing chess. This train of thought would further indicate that classifying a position as a certain opening activates certain long-term plans that you read/reinforcement-learned to be viable in these structures.
>
> Basically, positions and parts of positions remembered as chunks and calling these chunks "patterns", "pawn chains/islands/complexes" or "open files" and "attacking files" (I know implicitly that the rest of the file is emptied), and assigning these chunks to another level of chunks that we label "openings" might be one reason for the highly efficient recollectional abilities of highly rated and trained individuals that (at least partly) go away when being robbed of the "experience" that lays the foundation to this recollectional ability in standard conditions.
>
> So while I'd also consider chess recognition as quite a domain-specific ability, I think the outcome one can measure in these studies might not (and not be trying to, either) capture the ways in which the recollectional pathway might utilize many different complementing resources such as chunks, opening prototype positions, simple logic (I know this b4-pawn wasn't attacked, so this knight, which I'm unsure of where exactly it stood, can only have been on this square, etc.) and pure memorization, for exceptions/deviations from conventional and often-seen positions.
>
> Just my ramblings anyway, with anecdotal stuff sprinkled in between. Thoughts?
I agree with your last point about multiple aspects of chess recognition/memory. I don't think it's likely that there is some kind of unitary "chess recognition" mechanism, but instead multiple contributing processes that are still mostly limited in scope to chess. Some of what you're describing in the first part of your post makes me think of another topic I'm planning to cover soon: The Einstellung Effect. This refers to a sort of "false alarm" in pattern recognition that can lead experienced players to overlook an optimal solution because the position resembles something they know closely enough that they opt for the move suggested by that previous experience. I think this is a neat way to understand trade-offs between various kinds of grouping and chunking in perception and memory and more deliberate problem-solving.