ma𝕏pool on Nostr: nprofile1q…ufa4k nprofile1q…cvgna You can model different strategies as ...
nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpqknzsux7p6lzwzdedp3m8c3c92z0swzc0xyy5glvse58txj5e9ztqaufa4k (nprofile…fa4k) nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpq0l88929ze7h5h7g7pu7y0e0077c03zqtp9zk63zpx4gg7m2ypcnqmcvgna (nprofile…vgna)
You can model different strategies as population-level explore/exploit games. You have a multi-armed bandit but know only rewards from levers you have tried (and can exploit) or see someone else explore (and can copy) or you must explore yourself and try some new lever.
If animals are smart and live in packs, they can observe and learn from each other (copy). When someone scores everyone quickly learns new exploit, but only one pays the price if the pay-off is negative.
You can see this with humans. One pokes the thing and everyone else cheers them forward from a safe distance and observes. Probably the same with raccoons.
You can model different strategies as population-level explore/exploit games. You have a multi-armed bandit but know only rewards from levers you have tried (and can exploit) or see someone else explore (and can copy) or you must explore yourself and try some new lever.
If animals are smart and live in packs, they can observe and learn from each other (copy). When someone scores everyone quickly learns new exploit, but only one pays the price if the pay-off is negative.
You can see this with humans. One pokes the thing and everyone else cheers them forward from a safe distance and observes. Probably the same with raccoons.