“Well done, Mr. Armstrong”

As the late Neil Armstrong once said to the BBC’s Pallab Ghosh: “The dream remains! The reality has faded a bit, but it will come back, in time.”

I can’t watch either one of these historic moments again without thinking of the other, so I mashed them together.

About stevegrand
I'm an independent AI and artificial life researcher, interested in oodles and oodles of things but especially the brain. And chocolate. I like chocolate too.

10 Responses to “Well done, Mr. Armstrong”

  1. Mellowcow says:

    I don’t know what we’ll find out there but space sure is pretty. 😀


    • stevegrand says:

      Fantastic! So beautiful in their own right, and then when you think what you’re actually looking at…

      • Mellowcow says:

        My crude attempt to use it as a wallpaper.
        Anyway, don’t leave it up to space exploration to find new forms of life. I believe in you. >:O

      • stevegrand says:

        Cool. Ok if I use it on my desktop too?

      • Mellowcow says:

        Haha, sure. It’d be an honor. :3
        When I watched your video, I thought: “Where have I seen this before?”

        The Verge made this fake ad as a mockery of Microsoft’s overly dramatic marketing of their Surface. Ironically, it’s interesting and effective, at least to me. I want to go out and buy some Ovaltine right now!
        Okay, enough with this nonsense. See you around.

      • stevegrand says:

        Heh! So glad I didn’t set it to music now…

  2. Gryphon says:

    I’ve been trying for DAYS not to post this, but I can’t hold back any longer. A little bird flew by singing about how you may or may not have been musing about “play” and I’ve been frothing at the mouth quietly to myself ever since, though I’m aware it’s very naughty of me.

    At first I typed up like three pages going into my thought processes in detail but as we all know that’s only interesting when you do it. 😛 I’ll try to keep it short. My thoughts boiled down to this thesis statement: learning how to initiate play isn’t the baby’s problem. It’s the caretaker’s, whose imperative is to get the tiny thing to stop wailing as soon as possible. Babies are introduced to things that are fun and learn to enjoy them by being walked through the motions by their caretakers. Support for thesis: all the bad things that go on cognitively and behaviorally with babies NOT intensely trained by caretakers in how to play.

    Human babies are physically helpless, forcing them to interact intensively with their caregivers in order to meet their needs. I argue that getting things done by proxy in this fashion totally counts as tool use, and that having the first agent whose inputs and outputs you map be “another human being” jump-starts cognitive development LIKE WHOA. Learning to play “the caretaker game” provides an intense framework in which to learn things about causality and goal-directed behavior. Figuring out how to direct the caretaker’s attention in order to successfully get handed a distant object teaches a baby how to track multiple variables and the repercussions of its own actions at a distance–the fundamentals of understanding causality.

    And this all happens way pre-functional-mobility, due to that very lack of mobility. If babies were as in control of their own destinies as fawns or puppies, they would just learn how to walk over to things and play “how well can I predict the behavior of these physics objects?” games. But due to their physical limitations, bored babies must also play the vastly more complex “how well can I predict what my caretaker and I are going to do with this physics object?” game, where instead of the predicted end state being the relatively simple Well this block will probably topple when I push it, the predicted end state of a baby-caretaker-teaching-game interaction could be any completely arbitrary thing. Like, the difference between picking up and dropping a balled up piece of paper, and having an intensely social interaction with a caretaker in which the caretaker coos and smiles and tickles you only when you drop it through a tiny hoop, and gives you encouraging cues to keep trying otherwise, is HUGE. It’s like the difference between primary and tertiary protein structures, it’s orders of magnitude more complex and meaningful.

    I tried to get ^^^ to coherently support the argument that “babies need to be taught how to play meaningfully, see: neglected ones growing up all messed up,” but I couldn’t quite think of a graceful transition, so I’ll just say it. Like, real, goal-directed, intelligible play doesn’t arise spontaneously from innate curiosity and boredom drives such as the ones that govern a baby’s initial mastering of concepts like “The movements I can make seem to be under the constraint of my limbs not being permeable to one another.” Having fun, and learning to think of things to do that might be fun, (<—the most important part of generating meaningful play behavior, because actions can only be tried out once they occur to you, and how do you learn to have neat things occur to you when you're a blank slate, and the best times you've had so far have all been randomly twitching your limbs?) is something we are taught explicitly and at great length to do through interactions with other human beings.

    How simpler, in-the-now animals play seems to be mostly about getting their pattern generators set up right. They spend a great deal of time growing physically, and thus needing to constantly refine their understandings of how to pull off various physically complex but cognitively simple behaviors. Like pounce, tackle, change direction swiftly, figure out the immediate social inputs and outputs of other nearby organisms of the species, etc.. For human babies, though, I think the key difference is that from the get-go we intensely collaborate with a caregiver. This grasp of shared attention is both hugely important and completely inextricable from the development and execution of coherent, collaborative play–and even the later shape of solo play (see, children graduating from throwing and touching and banging blocks, to spontaneously doing abstract-goal-based things like gathering and stacking them). Caregivers, in constantly attempting to attract and direct the attention of babies, teach babies about what it means to attend, what it means to share attention, and how to differentiate self- from non-self input to objects. Through mimicry, they also impart to the baby various general rules about fun, how to play, and how to think of likely new things to master. I argued (uhhh well it's more like "arguing" now present tense ahahah oh well what is brevity) that the development of meaningful-to-an-observer play, the kind of coherent and sensible play of which one would ideally want the grandroids to be capable, does not arise intrinsically from in this case a tiny human's cognitive setup, but is learned the same way reading is. We've got the brainmap to be really good at it, but it's a social construct that does not/cannot arise spontaneously without input from other individuals. I feel like an A-life organism somehow wired to exhibit spontaneous, complex, iterative play entirely on its own would come up with some pretty dang inscrutable ideas about what constitutes "having a good time"…

    I also mused a little about how one might induce grandroids to play intelligently with each other and have some degree of preference for it over goofing off solo, but realized I had no idea what I was talking about. The only thing useful I can remember about that bit was "Boy, wouldn't it be nice to outsource some of this important social development to mature, caretaker grandroids, instead of risking every new one growing up with crippling cognitive disorders related to user neglect."

    wow ok a post about Neil Armstrong probably wasn't the best place to tack this on, and I'm still not entirely sure that posting this wasn't a really really REALLY rude thing to do, but the problems you face are SO INTERESTING I physically could not help myself. I promise I don't plan to make a habit of it, I have nothing but respect for the social order–I pledge allegiance, to the concept, that the Kickstarter elite are sacred; as are the forums, in which they post, forever and ever inviolate…

    • stevegrand says:

      Hahaha! You’re very welcome to post about Grandroids, and since there’s nowhere better, you don’t have much choice but to do it off-topic. In an ideal world, all the discussions about Grandroids would be out in the open, but I had to have something to offer the people who donated their hard-earned cash to the project and some exclusivity was almost all I had available. Also, this project is plenty hard enough without everybody and his dog wanting a say, so I need to keep it relatively close to my chest. But that only applies to people I don’t know.

      Interesting stuff! In truth your little bird misheard me somewhat, because I was stretching the word “play” way beyond its colloquial usage. I was asking myself and others what are the criteria for deciding when and how we experiment with the outputs of each cognitive module in order to give it some trials and errors from which to learn? We can learn to see (to some extent) simply by being in the world – things move past our eyes and things happen to us and we can passively learn to associate the one with the other. But when it comes to learning how to move our arm usefully, or how to walk, we can only do this if we actually wiggle the thing to get some feedback from it. It wasn’t clear to me when we might sit there (as newborn babies) and choose to flap our arms around, as opposed to some other part of our body, and when we do flap it, what is the actual structure of those movements, because they don’t seem to be purely random. As it stands, my creatures can only learn the rudiments of how to control their bodies by basically having epileptic fits, which at the very least doesn’t look natural. My mind was on extremely low-level behaviors – the kind of thing that would make dropping a paper ball through a hoop seem like an incredibly sophisticated action.

      But I totally agree that the feedback loop between a baby and its caretaker is crucial, at least in the real world. How much of that I can get into my game is a tricky question, because it requires the caretaker to be quite smart. Norns used to reflect off each other but because they were all equally stupid they’d just get into self-defeating spirals. But I completely agree with what you’re saying. Paradoxically, I’m kind of nervous about the interactions between creatures precisely because of the very truth of what you’re saying. We humans (at least, I presume most of my readers are human, although I think much the same would apply to many other mammals) are such finely honed social machines. The things parents spontaneously do with their children are perfectly tuned to enable them to learn the right things in the right sequence. There’s little chance that I can incorporate a lot of these instincts into my creatures so I’ve no idea whether they’ll just fail to learn or end up learning entirely meaningless things. I should pay some serious attention to this question, though, once I get the really primitive stuff out of the way. You wouldn’t believe what a challenge it is just getting them to avoid falling over, let alone having social interactions and recognizing what someone else is doing. But I need to at least keep it in mind when designing their visual system, etc. or I might miss some really important opportunities later. Thanks for the insights!

      • Gryphon says:

        Man, reading the things you write is like a breath of fresh air.

        >It wasn’t clear to me when we might sit there (as newborn babies) and choose to flap our arms around, as opposed to some other part of our body, and when we do flap it, what is the actual structure of those movements, because they don’t seem to be purely random. …

        They totally aren’t. I read an article once about some guys using bioluminescence to track the muscle movements newborn mice. In the early stages of normal development, the babies underwent frequent episodes of involuntary stretches, developmental seizures that gradually decreased in frequency as their bodies got wired up. I can’t for the life of me find the article but that’s probably because I’m using search terms like “glowing baby seizure mouse.”

        I really want to say something intelligent about biological pattern generators now but, again, I don’t actually have any idea what I’m talking about. I know spinal pattern generation is a thing that exists and is important, in its adult and refined form, for things like gait, but I mean…it seems reasonable to conclude, even based on that one study I read forever ago and only nebulously remember, that animals MUST come prepped with a set of such starting patterns! Clearly ones amenable to being overwritten and/or refined–see: infant grasp and smile reflexes, etc.–but definitely present.

        Which, is pretty obvious I guess actually, because clearly our respiratory and digestive systems come with factory settings. Swallowing, for instance, would definitely kill you to learn from scratch. What’s wrong with expanding the concept to an arbitrary set of starting motor imperatives? I mean foals rise to their feet because they are compelled to, with the same involuntary competence with which they avoid choking to death on milk.

        Of course, you don’t have any predators handy to drive the grandroids to ungulate levels of infant competence, but the infant mouse waves of muscular activity are somewhat relevant maybe? Gosh i wish I could find that article, they like, watched the signals cascade through the little wiggly mice and stuff, it was great. One of the few studies I’ve seen using luciferin/true bioluminescence! Which is miles better than pedestrian old fluorescence, as everyone knows. If I find it, I’ll post the link, if you’re interested, though I’m not sure if they crunched their data in any kind of well-tell-me-something-meaningful-about-the-morphology-of-the-signal-then way. I think they were just all “Oh hey guys so it turns out this is a thing that exists!” which, thanks, yes, but.

      • stevegrand says:

        Ooh, ooh, ooh!!! Gotta find the paper on the glowing mice! Maybe it doesn’t say much of any use but it could give some indirect clues for someone whose mind is already primed to see something. For years now I’ve been taking it for granted that there would be such spontaneous and yet not altogether random (i.e. coherent, wavelike) patterns of activity. I call them test-card waves, by analogy with the test-cards they used to broadcast in order to help people set up their analogue TVs. Stryker and Strickland’s work on how V1 organizes monocular and binocular cells relies on spontaneous seizure-like waves in the retina before birth and I’ve long thought that such a mechanism may be very general, and perhaps that the regularity of these strobing patterns may help the various coordinate frames of the brain’s maps build up. Years ago I was thinking about how such waves could act bidirectionally to produce a tension that morphs the coordinate frames into intermediates that connect the various hard-wired frames like retinotopic or homuncular. It was one of the key ideas that got me onto this new brain model in the first place but I’d kind of forgotten it!

        I should go back to thinking about it because it’s relevant to what I’m doing now, but it could also be critical for this play thing too, and I’m pretty excited about green glowing mice with epilepsy! Thanks for that! 🙂

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: