kitchen table math, the sequel: Jeff Hawkins TED talk

Sunday, August 12, 2012

Jeff Hawkins TED talk

Haven't watched it yet, but it's probably worth the time.

Here's the blurb:
Treo creator Jeff Hawkins urges us to take a new look at the brain -- to see it not as a fast processor, but as a memory system that stores and plays back experiences to help us predict, intelligently, what will happen next.

Jeff Hawkins pioneered the development of PDAs such as the Palm and Treo. Now he's trying to understand how the human brain really works, and adapt its method -- which he describes as a deep system for storing memory -- to create new kinds of computers and tools.

11 comments:

MagisterGreen said...

"You cannot learn or recall anything outside of a sequence."

That's my takeaway from this video. Very interesting stuff, and very much in line with a lot of stuff from people like Dan Willingham on how we learn.

I suppose now I have to go do some more reading on this. Need more hours in the day to learn stuff.

SteveH said...

If the goal is to find some science that will change the rote pedagogical beliefs of some ed school graduates, this won't do it.

Catherine Johnson said...

Need more hours in the day to learn stuff.

Yeah, well - you're way ahead of me!

I hadn't even finished watching the dang video!

I did discover, this weekend, that I have read nearly all of his book -- read and intensively annotated.

Now I need to read my underlinings and annotations.

In any event, I've held the 'brain is a prediction machine' view for ages now.

My VERY favorite 'science story' on this subject comes from Rodolfo Llinas, who describes a sea slug that .... (from memory) is a sea creature for a short period of time until it finds the spot where it's going to stay 'til it dies, at which point it eats its own brain because it no longer needs to make the predictions living animals must make in order to move.

Another factoid: forward motion is a POWERFUL trigger of 'ambition motions.' If you show college students an animation of abstract shapes moving 'forward,' they suddenly do better on whatever you've asked them to do.

Catherine Johnson said...

btw, I have a theory that as far as I know is unique to me.

Back when I was writing one of the Temple books, we got onto the subject of play. All animals play (I've forgotten now whether birds and fish play .... ) and, at least at that point, no one knew why.

Furthermore, as I recall, all animals play at the same stage of development & then stop playing at (roughly) the same stage of development. I think the timing coincided with development of the cerebellum.

There were various theories as to why, and one theory, which seemed to be getting a lot of play at the time, was that animals play in order to learn how to deal with 'surprise.'

So during a play-fight a dog will throw itself to the ground in order to learn how to deal with the surprise of an attacker knocking them off their feet, say. (That's the way I remember the argument.)

I don't buy that at all, mostly because it is not possible to surprise yourself, the same way it's not to tickle yourself. (Not being able to tickle yourself has to do with the cerebellum .... )

I think that play ***may*** be teaching animals not how to deal with surprise, but, rather, how to predict.

They go through all kinds of contortions and rolls and flops, and all of those actions teach them exactly what happens when they move (or get tossed) this way or that.

Moreover, since much play is done in pairs (or groups), they are also learning to predict the moves of their partner/slash/opponent.

Then, when their cerebellum is developed and they've taught it to predict movement, they stop playing.

It's a thought.

Catherine Johnson said...

Magister Green --- wow --- that statement about all memory being sequence ....

Pretty incredible when you think about how hostile the education establishment is to sequence.

Everything is supposed to be spiraled and/or taught in random thematic 'units.'

If all memory is sequence, and the sequences are recalled autoassociatively (which I believe....) ----

Boy, are we doing things wrong.

Glen said...

My own theory of play is that it is an evolved activation of the attentional mechanism that spurs you to expend energy and take risks to gather training data of all sorts. All that activity and exploration is costly--it gets some kids killed--but the up-front training is so valuable that it produces a net increase in likelihood that you will pass on your genes. At some point, the additional learning isn't worth the additional cost, and your attentions turn to other things.

Catherine Johnson said...

Offhand I don't see how your theory applies to animals -- ??

As far as I know, animals are never killed in play....I'm trying to think how often children die in play .... and I'm thinking that probably on the Serengetti plains dying in play would have been quite rare.

In other words, I'm thinking the built environment is what makes play dangerous to some kids ....

Glen said...

Children stick their hands in holes, get bitten, and die. They put everything in their mouths--dirt, rocks, leaves, flowers--not for food but for the experience. The sometimes choke or poison themselves. They try to break a branch with a sharp rock, cut themselves instead, and die from the infection. They reach out to try and touch a fish, fall in the river, and drown. They're so busy chasing each other on the Serengeti Plain that they don't notice the snake. They climb a tree, fall, and...you get the picture.

Children at play put themselves at risk of injury or predation (of the wild or urban type.) They burn up calories that are not necessarily abundant, putting themselves at further risk.

Hiding quietly in a hole with Mama conserving energy would be safer, so why play at all? Because, I believe, animals capable of learning (beyond instinct) have a higher expected value of passing on their genes if they get out and play (train their brains and nervous systems) than if they don't.

Allison said...

--As far as I know, animals are never killed in play..I'm trying to think how often children die in play ....

Well, certainly animals die in play if it interferes with their noticing of a predator or deadly situation. Cavorting in water that you don't realize has a croc in it will get you killed. squirrels die from falls at times--surely some of those jumps are play.

Kids drown during play quite a lot here in MN in the summer. no built environment needed for that to happen.

i don't actually see much daylight between you and Glen on this, catherine. human and animal traits evolved for the same reasons, that those traits oinferred advantage. but I'd add that in addition to play being training and risking, it also showing fitness in itself. if a creature can afford to play, much like if a peacock can afford to look good, then it must be very fit, with health and energy to spare.

Anonymous said...

My felines and canines are 15, 14, 13, 5, and 5, and they all still play.

AmyP said...

There's been a lot of exposure given recently to a National Geographic study done with kitty-cams to figure out what cats get up to on their outdoor rambles. As you might expect, there's a lot of killing of small animals (for instance lizards), and relatively little eating of those small animals. I think that we can probably categorize the killing not followed by eating as play or practice. It keeps the kitties in fighting trim in preparation for the possibility of having to fend for themselves.