Why A Good Memory Is Bad For You | Hacker News Reader

Why A Good Memory Is Bad For You

technologyreview.com

34 pointsca98am7915 comments

Couldn't foragers remember food sources perfectly well, but still not exclusively visit them? I don't see how these results necessarily implicate memory. Habit, sure, but not memory.

Quite agree, this paper (or at least the summary) is nonsense.

And this:

  Take as an example the question: if the ability to
  remember is such a good thing, why hasn't evolution
  given us photographic memories? The answer according to
  Boyer and Walsh's model is because we'd starve when the
  local supermarket went out of business.

Has to be the stupidest thing I can ever remember reading in a purportedly scientific article. If I have photographic memory why would I visit the supermarket more than once, after it closes? Just because I remember that there was food there doesn't mean I'm required to go there.

It may be nonsense when applied to humans with our fancy pre-frontal cortices, but in lower animals there's no interdiction between memory and action—that is, animals always do "the first thing that comes to mind." To put it another way, every memory an animal has is a habit. You would visit the supermarket after it has closed because it's hard to break that habit (disestablish a memory with a number of positive re-trainings), even when you "know" that it's no longer true (from a single negative re-training).

Consciousness is theorized to be precisely the brain experiencing indecision between two alternative motor programs (i.e. habits), and rationally picking one or the other. Animals don't do that.

> Animals don't do that.

I hope you don't really believe this. Animals are not automatons, they do think and they do plan.

Animals may plan sometimes, but it seems to be the exception, not the rule.

"According to the 'mental time travel hypothesis' animals, unlike humans, cannot mentally travel backwards in time to recollect specific past events (episodic memory) or forwards to anticipate future needs (future planning). Until recently, there was little evidence in animals for either ability. Experiments on memory in food-caching birds, however, question this assumption by showing that western scrub-jays form integrated, flexible, trial-unique memories of what they hid, where and when. Moreover, these birds can adjust their caching behaviour in anticipation of future needs. We suggest that some animals have elements of both episodic-like memory and future planning."

source: http://www.nature.com/nrn/journal/v4/n8/abs/nrn1180.html

I never said animals didn't plan, and make decisions; I said they weren't conscious. There's a big difference.

Planning is the evaluation of alternative, non-conflicting motor programs to determine the cumulative utility of a decision branch. (Basically, a decision-tree A* algorithm.) Animals do plan, but planning is not experienced consciously—it simply happens.

Human planning happens on this pre-conscious level as well. If you've ever been in a "flow state" where you felt like you didn't actually have to think about anything, because you just "reacted" the right way to every problem—that's planning. That's how it feels to be an animal.

Thinking is, as far as neurology has figured, the resolution of problems that occur during planning by the use of the neocortex to classify features of the current environment into schema, and re-trigger the planning process with expectation programs, looked up from those schema, substituted for motor programs. The most common form of thinking involves verbal expectation programs (basically, saying things to ourselves, in our heads, and then reprocessing those words as if they were external sensory inputs.) This is experienced as consciousness, and as far as we know, only humans do it.

Without the ability to think consciously—to resolve motor-program conflicts—an animal can't "argue" with itself about which course of action to choose. Instead, it just has a best intuition about which one will work out, and goes with it. These intuitions rely greatly on habituation—in fact, without expectation programs (and thus without semantic knowledge), memory as a whole consists purely of process knowledge—internal finite-state machines of habituated reactions.

And all that somehow implies they won't forage until they forget where they've found food before? Even if they've since struck out at those same locations?

> This is experienced as consciousness, and as far as we know, only humans do it.

That's simply entirely not true. As far as I know, the only ability uniquely found in humans is the ability to think about thinking, everything else we do can be found in the animal kingdom.

Consciousness is not a binary state, there are many levels of it and while we sit at the top of the pyramid, it is not a feature unique to us.

> if the ability to remember is such a good thing, why hasn't evolution given us photographic memories?

And the other answer: Evolution doesn't select for perfection, just for sufficiency. Even in the context of this article, in terms of a survival, a photographic memory is overkill.

One possible answer is that yes, that would be a good way to design a brain, but no we can't because evolution isn't smart. So we have brains that overwhelmingly lock on to whatever behaviors we remember being rewarded in the past, and only go exploring when we don't find food other places we remember (which may be too late)

It may be 'easy for evolution' to limit memory, but 'hard' (more unlikely mutations on a more delicate structure) to make a planner that revisits known food sources only a limited amount.

So a forager makes a fruitless (har!) trip to the apple tree, explores elsewhere, and finds food. That slightly deconditions the "revisit" behavior, and slightly conditions the "explore" behavior. Such a planner doesn't seem too difficult for evolution to generate.

But do you know for a fact that searching the apple tree every day until failure, and then traveling elsewhere, is any better and in fact not worse than a combination of randomness and memory?

After all, one of the first things you learn in hunter-prey environments (capture the flag, online fps, etc) is you want to be somewhat unpredictable, whether you are hunter or prey. When you are always in the same place, it decreases your survival chances. Considering wild animals are ALWAYS in a hunter-prey environment, it doesn't seem a stretch that this randomness might benefit them in this fashion.

I'll have to remember this next time I'm foraging at Whole Foods.

Evolution probably hasn't given us a photographic memory, most likely, because doing so doesn't provide a benefit worth the expense. which is not the same as saying photographic memory would actually hurt us.

I think its not only about memory, but about how rewarding it is to seek new opportunities than constantly sow the existing ones.

Reminds me of Metropolis light transport.

Reply on news.ycombinator.com