Imagine the simple task of finding words in a wordsearch.
Initially the default strategy might be just to let your eyes wander around to see if you spot any words.
Strategy 2 might be to search for the first letter of a word.
Other strategies might be to go line by line to see if you see a word.
Now, my question is how are these strategies learned (or invented?).
My intutition is that we can only make very small leaps of reasoning from something that we do already. So for example the line-by-line strategy might not be a huge leap from the strategy that we already have in reading text line by line.
Or I expect we already have some mechanics to follow paths, so we can build on that.
The random-search strategy probably is inbuilt in us, used for things like looking for fruit in a tree.
I'm pretty sure I "invented" a line-by-line strategy myself without being taught it. Round about ages 6-8. But I'm not sure how I invented this algorithm or proved to myself that this would indeed be garunteed to find an answer. (If the brain was just coming up with random algorithms, then many would be useless. Perhaps it does and then we self-analyse those strategies, e.g. "this strategy goes over the same letters", "this strategy misses out a lot of letters" until it hits on a good one.)
So how do we come up with new strategies for things, and is linguistic based reasoning important for this? I feel like if we could work out how strategies are invented this would help build a model of the brain or a general AI. The wordsearch example is the simplest example I could think of which is non-trivial yet not too difficult.
(I note that the task of "find words in this grid" has to be communicated linguistically in the first place).