Chunking effects on the n-back task
Chunking effects on the n-back task
User cevapcici has written in that user DoubleT has gotten 100% on the n-back task with n=9, while cevapcici has reached n=8 and has begun to experience chunking effects.
What does this mean, and why is it suboptimal? Chunking refers to when the brain begins to manipulate groups of objects as a single unit. For example, suppose you see the number 63463451. Instead of processing this as a plain 8-digit number, your brain is likely to remember "634", twice, and "51," which means you can recall 8-digits while only using the space for 5, plus a "repeat this" command.
Chunking happens when there is a familiar pattern that can be accessed quickly, which can come from both short-term/working memory and/or long-term memory. For a long-term memory example, if you were to see 314127181414, you might recognize that this is the first 4 digits of pi, e, and the square root of 2 put together. Then, you would only need to remember 3 pieces of information (plus their positions) to recall the whole 12-digit string! Another example would be 7654321, or your phone number.
If chunking happens on the n-back task, that means you have become familiar enough with certain sequences of items, like "left-top, right-center, left-center" or "C G A" (maybe even "middle-bottom-K, right-top-T"), such that you are thinking of them as one piece. This also means the cognitive load required for the task has decreased, and your brain is no longer "learning" as much. If you are chunking 3 items together on the 9-back task, it might be as "easy" as it was when you just begun the 4-back task.
What does this mean, and why is it suboptimal? Chunking refers to when the brain begins to manipulate groups of objects as a single unit. For example, suppose you see the number 63463451. Instead of processing this as a plain 8-digit number, your brain is likely to remember "634", twice, and "51," which means you can recall 8-digits while only using the space for 5, plus a "repeat this" command.
Chunking happens when there is a familiar pattern that can be accessed quickly, which can come from both short-term/working memory and/or long-term memory. For a long-term memory example, if you were to see 314127181414, you might recognize that this is the first 4 digits of pi, e, and the square root of 2 put together. Then, you would only need to remember 3 pieces of information (plus their positions) to recall the whole 12-digit string! Another example would be 7654321, or your phone number.
If chunking happens on the n-back task, that means you have become familiar enough with certain sequences of items, like "left-top, right-center, left-center" or "C G A" (maybe even "middle-bottom-K, right-top-T"), such that you are thinking of them as one piece. This also means the cognitive load required for the task has decreased, and your brain is no longer "learning" as much. If you are chunking 3 items together on the 9-back task, it might be as "easy" as it was when you just begun the 4-back task.
cognitivefun | 4 years ago
Reply
How does a person sense the danger of developing task-specific strategies?
I found myself remembering a sequence and repeating it in my head. I noticed this helps my score compared to rounds when I make no direct attempt at remembering sequences and just recall with some historic sense of familiarity, however I'm concerned whether such attempts at memorization are damaging to the 'augmentation' process.
To generalize the question, is it best to play with a blank mind and no strategy?
I found myself remembering a sequence and repeating it in my head. I noticed this helps my score compared to rounds when I make no direct attempt at remembering sequences and just recall with some historic sense of familiarity, however I'm concerned whether such attempts at memorization are damaging to the 'augmentation' process.
To generalize the question, is it best to play with a blank mind and no strategy?
rectotron | 4 years ago
Reply
"To generalize the question, is it best to play with a blank mind and no strategy?"
I am not sure if there is a "best" way of going about these tasks. You can probably find a dozen people using a dozen different strategies, but the only things that are unambiguously measurable and comparable are the numbers. There may be efficient or inefficient strategies, but the end goal is all you're looking at.
Speaking of mnemonics though, to make this relevant to your question, I would say it would undermine the task's principle if you started devising mnemonics for these stimuli. For example, if you invented a mnemonic to map different consonants to each of the 8 visual stimulus positions and connect them with vowels as you go. As long as you don't go this far I would assume you are practicing something relevant.
Speaking of mnemonics though, to make this relevant to your question, I would say it would undermine the task's principle if you started devising mnemonics for these stimuli. For example, if you invented a mnemonic to map different consonants to each of the 8 visual stimulus positions and connect them with vowels as you go. As long as you don't go this far I would assume you are practicing something relevant.
cognitivefun | 4 years ago
Reply
"How does a person sense the danger of developing task-specific strategies?"
I think that is a difficult question, because it's very hard to keep your brain from committing things to memory and taking the shortcut of retrieving from it to speed up pattern recognition -- not that this is undesirable, but it would lessen the focus on the novel pattern detection and increase the focus on memory retrieval.
While I wouldn't call it a danger, I'm sure many people are wondering to what extent are cognitive drills domain-specific. While Jaeggi and Buschkuehl et al.'s research proposes a certain degree of transferability from their dual n-back task, it is far from conclusive.
On the other hand, if you getting better at developing strategies, that itself may be a useful skill.
While I wouldn't call it a danger, I'm sure many people are wondering to what extent are cognitive drills domain-specific. While Jaeggi and Buschkuehl et al.'s research proposes a certain degree of transferability from their dual n-back task, it is far from conclusive.
On the other hand, if you getting better at developing strategies, that itself may be a useful skill.
cognitivefun | 4 years ago
Reply
I have only recently scored about 80% on 3-back, so maybe someone else has a more interesting strategy, but I was thinking about the same thing. I have tried both remembering the sequence, and relying on vague familiarity. Only in the former case am I able to get anywhere.
I also tried sequentially knocking out the last item, as the new item moves in, but this doesn't seem to work either, at least for me. I end up keeping fixed slots and replacing them accordingly. But this replacement seems to be getting easier.
I also tried sequentially knocking out the last item, as the new item moves in, but this doesn't seem to work either, at least for me. I end up keeping fixed slots and replacing them accordingly. But this replacement seems to be getting easier.
caffeinecookie | 4 years ago
Reply
"I also tried sequentially knocking out the last item, as the new item moves in, but this doesn't seem to work either"
That's the first thing I tried, what a disaster :)
Remembering a sequence for the first few instances and then recalling for the next few works fairly well, but the segmentation assures you're going to lose the sequence during the 'recall' phase and have to rely on old fashioned luck that nothing comes up during the 'remembering' phase. It's a safe middle ground. While it prevents complete failure, it also prevents results that are anything to write home about. Plus this approach certainly fits the "task specific" category, so it defeats the whole purpose.
That's the first thing I tried, what a disaster :)
Remembering a sequence for the first few instances and then recalling for the next few works fairly well, but the segmentation assures you're going to lose the sequence during the 'recall' phase and have to rely on old fashioned luck that nothing comes up during the 'remembering' phase. It's a safe middle ground. While it prevents complete failure, it also prevents results that are anything to write home about. Plus this approach certainly fits the "task specific" category, so it defeats the whole purpose.
rectotron | 4 years ago
Reply
