Dave’s Whiteboard

Chris Chatham at Developing Intelligence reports a study in which a greater amount of working memory seems to hamper learning rather than aid it.

The link takes you to the full post, but here’s one of the highlights:

…subjects who were able to remember more letters [in the experiment's tasks] actually took longer to learn the complex categorization rule! (Here, “learning” is defined as the number of trials until they got 8 trials correct consecutively). This result was reversed for the simpler rule, such that subjects with better working memory learned that rule more quickly. Furthermore, memory scores were unrelated to perceived pressure, providing preliminary evidence that higher scores are not merely related to motivation.

“Simple categorization” in this case was classifying objects according to one dimension — e.g., if the dimension were shape, you might identify all the circular things. “Complex categorization” would involve two out of three factors — e.g., squareness, redness, largeness.

Here’s a related post by Kumar Narayanan summarizing some thinking about working memory.

In comments to this post (about women in the “edublogosphere”) at Janey Clarey’s blog, Karyn Romeis talked about what I’ll call perceived expertise. The organizer of a tech conference asked why she wasn’t giving one of the seven-minute micropresentations or two-minute nanopresentations.

Karyn said in part, “…I didn’t feel that I knew enough about anything to stand up and tell other people about it…”

I know that feeling. I’ve known it most of my career.

A workshop I attended early on introduced me to lean programming and other ideas developed by people like Dale Brethower and Geary Rummler. That in turn led to my first ISPI conference, where the variety of talent and the openness of sharing were intoxicating.

(Something like Winston Churchill talking about FDR: “Meeting Franklin Roosevelt was like opening your first bottle of champagne; knowing him was like drinking it.”)

For a long time, I felt as though others had things to teach, and I had things to learn. My project at the time was creating computer-based training for Amtrak’s new reservation system. I was so focused on the specifics of our software that I couldn’t see how anyone would be interested in whatever I was doing.

But I wanted to participate in that community of practice, to contribute as well as benefit. Stepping back from the details of our CBT package, I thought about our team’s successes and what had led to them. We’d used a lean, inductive approach; we created short, focused courses; we avoided pointless midlevel branching; we had a high level of interaction; and we supplemented formal instruction with the creation of “training trains” on which people could practice any transactions of the actual Amtrak system.

And we learned a lot from trying things. “Good judgment comes from experience; experience comes from bad judgment.”

I decided I didn’t have to be the ultimate guru regarding CBT, which in those days was slightly less confusing than the Code of Hammurabi. I could share what we at Amtrak had done, why we did it, what worked, and what didn’t.

I would go with what I knew, without pretendingI knew everything.  This was the description in the conference brochure:

CBT: Your Mileage May Vary

This session presents ways that CBT offers individualization and student feedback and asks: why bother? It keeps the delivery cart behind the instructional design horse. You’ll find out what you could do, ways to decide, and examples of things to keep in mind when looking at CBT systems. The presentation includes mistakes we made. Based on Amtrak’s experience with the Scholar/Teach 3 system, but requires no detailed CBT knowledge.

I’d love to be a guru (or even been seen as one, though that’s a different issue), but I’m not. What works for me, and what I encourage others to consider, is to talk about “here’s what I’ve found” rather than “here’s what I know” (or, heaven help us, “here’s what you should know”).

(Thanks to Karyn for agreeing to be quoted in this post — and for triggering it in the first place.)

The Cognition and Language Lab has its own blog, where I found Why Languages Can’t Be Learned.

That post in turn led to an experiments link on the CLL site, where you can participate in experiments (typical completion time: 5 minutes).

The original trigger for me was this Chris Chatham post at Developing Intelligence. He includes a link to his own 2006 post on machine and human learning of word meanings.

That last post describes Latent Semantic Analysis, a set of algorithms assessing the similarity of words to other words. I especially liked this conclusion:

After training, LSA performed at 64.4% correct on a multiple choice test of synonymity taken from TOEFL (in contrast, humans score around 64.5% on average on this test, which is frequently used as a college entrance examination of English proficiency in non-native speakers. By this metric, LSA would be admitted to many major universities!)

Finally, I came across a link for this terrrific video. I couldn’t remember its name (Le Grand Content) or its director (Clemens Kogler), making it awfully hard to search for. (Also available at his site.)

One more reason to use del.icio.us.

I’ve reread two of Steven Pinker’s books — How the Mind Works and The Blank Slate — and look forward to doing so again, both for the quality of the writing and the stream of ideas.

Thanks to Amy Gahran at Contentious.com, I found a Steven Pinker multimedia page featuring many media files — some audio, some video…

Like Pinker’s TED talk entitled The Stuff of Thought: