Basal ganglia to the rescue (or how to walk in Paris)

In France this summer, I had an illustrative experience re:information-integration learning.”

Information-integration” learning is unconscious or implicit learning (heavily reliant on the basal ganglia). It’s the opposite, more or less, of book learning. 

I say “more or less” because I’m not sure the two systems are ever completely separate.

They seem to specialize in different jobs; that much, I think, is clear.

They seem also to compete: when one is “turned on,” the other is “turned off” or at least “turned down.” (I think that’s right–I remember reading it–but I don’t remember my source. So take it with a grain of salt.)

In any event, it appears that people really can be better at one than the other, as folk psychology would have it. There are intuitive types, and there are cerebral types. 

But the two systems seem to support each other as well, though I don’t think anyone knows how that works. I’m not sure anyone even has a proper theory, though I could have missed it.

Anyway, back to my summer.

You do a lot of walking in Paris. Obviously. (Sidebar: that new study finding that exercise, in the form of brisk walking, does produce weight loss if you do enough of it certainly holds true for me.)

The problem with walking a lot in Paris is that there are no flat surfaces anywhere, ever. Every sidewalk is a zillion years old, and no sidewalk has ever been re-finished or re-poured or whatever it is people do to ancient sidewalks to make them walkable and safe.

Not to mention all the Parisian dual mini-curbs. You’ll be walking along, and, all of a sudden and out of the blue: CURB.

CURB, and not where you’d expect it, either, i.e. not at a discernible corner, where a proper curb ought to be.

Then, 1 step further along, SECOND CURB. Double curbs in the middle of a block!

Why?

Early in our stay, walking down a broad sidewalk in the 14th, I think it was, staring at my Google map trying to figure out where the h-e-double-hockey-sticks I was, I took a really bad fall–straight onto my knee, hard, then onto my wrist. Bam, then bam again. 1

After that, I kept taking multiple almost-falls: step, step, trip, lurch, catchmybalancestep, step, trip, lurch, catchmy-balance, repeat, repeat, repeat. Incredibly not fun. I wasn’t having fun anyway (that’s another story–and, yes, I know it’s a Life Failure not to have fun in Paris) and the constant tripping-lurching-balance-catching added to the woe.

I was beginning to think the problem was that I’m old, and was spending time wondering how any old people ever manage to walk in Paris, or do they all just drop dead of broken hips at 70 or 80, when I remembered that the one and only time I’ve ever sprained an ankle happened in Paris, as I stepped off a curb, at age 30.

Which of course raised the question of why I wasn’t seeing 30-year old French people falling off curbs, but never mind. 

Point is, I was not enjoying my hikes across the city or even my shorter expeditions in the neighborhood (which was fantastic, by the way! Rue Daguerre. If you’re going to Paris, stay in the 14th!)

I’d had one bad fall, and the constant near-misses were driving me crazy.

Then one day I looked up from Google maps and realized I had been sailing down the Boulevard Raspail for blocks and blocks without having once checked the sidewalk for potholes, bulges, or double-curbs in the middle of nowhere. I was just walking, and walking fast to boot, without looking down. 

I was walking like I lived there–which, by then, I basically did. 

That has to have been information-integration learning. Those weeks (two or three weeks, maybe?) of trial-and-error walking taught my unconscious brain how to walk in Paris and not fall down. 

I couldn’t begin to tell you what I was doing differently.

Was I lifting my feet higher than I do in New York?

But then wouldn’t that make the potholes more of a problem? 

Was I checking the sidewalk without realizing it?

I have no idea.

It was pretty great, though.

Information-integration learning Rule-based learning 
“Life” learning “School” learning
Unconscious Conscious
Implicit learning: you can’t necessarily put what you’ve learned in words (& if you can, words come to you later) Explicit learning: you can put what you’ve learned in words
Intuition, everyday categories (good versus bad, dog versus cat), social rules, habit Formal concepts, theories, disciplines, etc.
Grammar Vocabulary
Learns relatively slowly Learns quickly
Can’t learn “offline” (learning stops after a “lesson” is over) Can learn “offline” (learning continues after a lesson is over)
Must have immediate feedback – students must know whether their answer was right or wrong after each answer or no learning occurs Can learn with delayed feedback – students can get their tests back days later and still learn from their mistakes)
Can learn several things at the same time (e.g.: can learn the orientation and the width of a stimulus) Can learn just one thing at a time (can learn the orientation or the width of a stimulus, but not both at the same time)

Source: Deferred Feedback Sharply Dissociates Implicit and Explicit Category Learning
J. David Smith, Joseph Boomer, Alexandria C. Zakrzewski, Jessica L. Roeder, Barbara A. Church, and F. Gregory Ashby  Psychological Science 2014, Vol. 25(2) 447-457

And see:
The most important research on learning I’ve ever read 

1. For the record, this is why I don’t get bone scans. Every 3 or 4 years I take a really hard fall, usually dog-related, I land full-weight on one kneecap or the other–or, alternatively, half-weight on one kneecap & half-weight on one wrist–then I get up dazed but intact. Dazed-but-intact after a headlong fall onto a hard surface is a diagnostic, I figure. 

gasstationwithoutpumps on the two learning systems and grammar

Succinct and on the money:

Big chunks of grammar are rule-based learning, at least at the level of what distinguishes academic writing from casual conversation. The rules are articulated in grammar handbooks and can be consciously applied.

Grammar at the level of what sentences one can use in casual conversation is much more “information integration”, as it takes skilled linguists substantial effort to express the grammatical constraints in rules, and fairly complicated rule systems are needed for even crude approximations to grammaticality

That’s exactly right.

The principles Katharine and I teach in our curriculum can be learned–quickly learned–via rule-based learning:

  • End focus: put the most important information in the sentence last 
  • Known-new contract: start with information the reader already knows, proceed to new information he or she doesn’t know (or hasn’t heard you say yet)
  • Cohesive topic chains: many if not most of your sentences in a paragraph should have the same or closely-related grammatical subject (I think the most effective percentage in a fairly long paragraph is around 75%)

And see:
The most important research on learning I’ve read

The most important research on learning I’ve read

Someone has posted it on line, so here it is !

Deferred Feedback Sharply Dissociates Implicit and Explicit Category Learning
J. David Smith, Joseph Boomer, Alexandria C. Zakrzewski, Jessica L. Roeder, Barbara A. Church, and F. Gregory Ashby  Psychological Science 2014, Vol. 25(2) 447-457

I’ve been mulling this article since the summer of 2014, when it was published. It’s life altering.

Unfortunately, if you’re not in the field, reading the study isn’t easy. The press release is intelligible but brief. 

For the moment, the thing to know is that this study probably proves, finally, that we have two separate and distinct learning systems inside our brains.

Psychologists and cognitive scientists have been talking about “dual systems” and “dual systems theory” forever, but no one had nailed it down. Now they have.

The two systems are separate and distinct in the sense that if you turn one system “off” the other still learns. They’re “dissociable.” 

And: the 2 systems learn differently.

“Deferred feedback” looks at category-learning, but as far as I can tell these two systems learn everything, including physical skills. 
~

Compare and contrast

The chart below is correct (I believe), but it says nothing about the relationship between the two learning systems–which no one seems to understand yet. 

So, while I’ve put “vocabulary” under explicit learning, I’m fairly sure vocabulary can also be picked up via implicit learning. 

And given what I’ve seen in the L2 literature about grammar learning, it seems clear that some explicit learning helps with grammar, too–at least, with the kind of grammar you use in formal writing, as well as with learning the grammar of a second language.

Obviously, no one learns the conversational grammar of his or her native language at school. 

In short, the two systems seem both to compete and to support each other in some way no one has worked out.
~

Information-integration learning Rule-based learning 
“Life” learning “School” learning
Unconscious Conscious
Implicit learning: you can’t necessarily put what you’ve learned in words (& if you can, words come to you later) Explicit learning: you can put what you’ve learned in words
Intuition, everyday categories (good versus bad, dog versus cat), social rules, habit Formal concepts, theories, disciplines, etc.
Grammar Vocabulary
Learns relatively slowly Learns quickly
Can’t learn “offline” (learning stops after a “lesson” is over) Can learn “offline” (learning continues after a lesson is over)
Must have immediate feedback – students must know whether their answer was right or wrong after each answer or no learning occurs Can learn with delayed feedback – students can get their tests back days later and still learn from their mistakes)
Can learn several things at the same time (e.g.: can learn the orientation and the width of a stimulus) Can learn just one thing at a time (can learn the orientation or the width of a stimulus, but not both at the same time)

~
From the abstract:

Deferred reinforcement qualitatively eliminated implicit, information-integration category learning. It left intact explicit, rule-based category learning.

Laurel, meet Yanny

In case you’re wondering, I heard “Laurel.”

Then I walked across the room and heard Yanny.

Which reminds me: I need to spend some time at Phonetique. I haven’t done so because, unfortunately, they’ve got their exercises set up wrong for implicit learning

Implicit learning (“information integration” learning) requires immediate feedback. You can’t do 10 items then find out which ones you got wrong. You have to do one at a time. 

Speaking of immediate feedback, this is the most important research on learning I’ve ever read.