Tuesday, August 25, 2009

Perception

Atul Gawande's article, "The Itch" is an interesting look at the mechanism of itching. My favorite part, though, is this:
A new scientific understanding of perception has emerged in the past few decades, and it has overturned classical, centuries-long beliefs about how our brains work—though it has apparently not penetrated the medical world yet. The old understanding of perception is what neuroscientists call “the na├»ve view,” and it is the view that most people, in or out of medicine, still have. We’re inclined to think that people normally perceive things in the world directly. We believe that the hardness of a rock, the coldness of an ice cube, the itchiness of a sweater are picked up by our nerve endings, transmitted through the spinal cord like a message through a wire, and decoded by the brain.
In a 1710 “Treatise Concerning the Principles of Human Knowledge,” the Irish philosopher George Berkeley objected to this view. We do not know the world of objects, he argued; we know only our mental ideas of objects. “Light and colours, heat and cold, extension and figures—in a word, the things we see and feel—what are they but so many sensations, notions, ideas?” Indeed, he concluded, the objects of the world are likely just inventions of the mind...
Although the article goes on to make it clear that this isn't complete, doesn't it sound like perception is just based on mental models?

I've been thinking a lot about simulations lately, and this article makes it seem like even fairly poor simulations could be surprisingly realistic if they abstract the right things or if people are able to get past the perception that what they're experiencing is a simulation:
The images in our mind are extraordinarily rich. We can tell if something is liquid or solid, heavy or light, dead or alive. But the information we work from is poor—a distorted, two-dimensional transmission with entire spots missing. So the mind fills in most of the picture. You can get a sense of this from brain-anatomy studies. If visual sensations were primarily received rather than constructed by the brain, you’d expect that most of the fibres going to the brain’s primary visual cortex would come from the retina. Instead, scientists have found that only twenty per cent do; eighty per cent come downward from regions of the brain governing functions like memory. Richard Gregory, a prominent British neuropsychologist, estimates that visual perception is more than ninety per cent memory and less than ten per cent sensory nerve signals...
The account of perception that’s starting to emerge is what we might call the “brain’s best guess” theory of perception: perception is the brain’s best guess about what is happening in the outside world. The mind integrates scattered, weak, rudimentary signals from a variety of sensory channels, information from past experiences, and hard-wired processes, and produces a sensory experience full of brain-provided color, sound, texture, and meaning. We see a friendly yellow Labrador bounding behind a picket fence not because that is the transmission we receive but because this is the perception our weaver-brain assembles as its best hypothesis of what is out there from the slivers of information we get. Perception is inference.
The article goes on to talk about the use of mirrors for phantom limb syndrome and other more interesting things. The idea is to reset the brain, so it stops thinking it is getting sensations that it isn't.

As always, I am thrilled when (a) computer science shows up in real life and (b) I guess right:
Researchers at the University of Manchester, in England, have gone a step beyond mirrors and fashioned an immersive virtual-reality system for treating patients with phantom-limb pain. Detectors transpose movement of real limbs into a virtual world where patients feel they are actually moving, stretching, even playing a ballgame. So far, five patients have tried the system, and they have all experienced a reduction in pain. Whether those results will last has yet to be established. But the approach raises the possibility of designing similar systems to help patients with other sensor syndromes. How, one wonders, would someone with chronic back pain fare in a virtual world? The Manchester study suggests that there may be many ways to fight our phantoms.
I wish the article was accessible to my students; it would be interesting to talk to them about perception and simulation. I might still try, or use the article for differentiation purposes.

Saturday, August 8, 2009

Relinquishing tight control

The link to "Death to the Syllabus" is going around the twitterverse. I must admit that the first third of the essay did not fully engage me; though I'm not sure why I persevered to read the whole thing, I'm glad I did because the last third was excellent enough to comment on.

Mano Singham makes a case that the traditional college syllabus, full of specific rules and consequences for tiny infractions, is a Very Bad Idea. "The implicit message of the modern course syllabus is that the student will not do anything unless bribed by grades or forced by threats." He goes on to mention,
There is a vast research literature on the topic of motivation to learn, and one finding screams out loud and clear: controlling environments have been shown consistently to reduce people’s interest in whatever they are doing, even when they are doing things that would be highly motivating in other contexts.
He laments that there is a negative cycle between students and teachers, where teachers do not feel comfortable making judgements about students' performance and behavior, where they instead create new rules to handle each situation. He mentions that making individual judgement calls is time-consuming and that in our legalistic society, teachers may feel defensive about making individualized decisions.

In the final section, he describes an experiment he's been performing in his courses for the past several years, where he gives a very open-ended syllabus, develops a classroom culture, then asks the students to create their own policies. He finds that students entrust his judgement and for the most part decline to make many specific rules.

It should not be surprising that I feel validated and encouraged by this essay. Without grades at my school, we rely largely on rubric scoring. (I recently discovered that the word "rubric" can have many meanings, what I'm thinking of is pictured here.) My rubrics tend to be vague, when I even make them at all. I do give students the list of criteria I'm looking at - algorithmic complexity, creativity, good documentation, whatever. But I hate rubrics that are overly specific, mostly because I hate grading that way. An example:

On our website rubric, one of the things we look for is good writing. Some of that is web-appropriate writing, like shortish paragraphs and clear sentences. Some of it is just plain-old good writing: correct spelling, good grammar, that kind of thing. Example rubrics are "specific." 0-1 misspellings will get you an A. 2-3 misspellings will get you a B. 4-5 a C, and so forth. Can I tell you how interested I am in spending quality time hunting down every misspelled word in a website and counting them? For every student? And then there's a whole list of other criteria to look for. I have better things to do with my time than count misspelled words. Especially since I'll then have to cross-reference with the list of kids who have accommodations for learning differences and can't be expected to spell correctly. And the kids who write more - which usually means a better project - will be penalized because they have more opportunities for misspellings. (Oh look, I got started on this. Aren't you lucky?)

Instead of putting specific thresholds of how many misspelled words, I tell the kids that spelling and grammar count, and if the spelling and grammar are bad enough to interfere with the quality of the project, they get dinged. But the rubric looks like, "Excellent grammar and spelling", "Good grammar and spelling", "Poor grammar and spelling", not specific numbers.

On the one hand, I understand that the common thinking is that students want specific guidelines. And yet, so much of life is in vague judgements. The kids know what excellent grammar looks like. They know what poor grammar looks like. They know how to get from poor to good (find an editor, use software tools...) and I don't think that any of our lives are enhanced by suggesting that in a whole website the difference between an A project and a B project is one misspelled word.

I think that by being vague in this way, and then being willing to engage in discussions with the students if they disagree with our assessments, we help students develop their own judgements. They should have a sense of what good grammar looks like, one they can apply without a teacher telling them if they're right. (Oh right, I'm a computer science teacher, not an English teacher. They should have a sense of what a well-documented, neatly coded program looks like, how about that?) Being open to the discussion is an important corollary - students should be able to question, teachers should have reasons, and teachers should be open to being convinced, though not too open.

The place where I'm less secure, but unlikely to change, is that I - like Mano Singham, I think - have a holistic view of assessment. I know teachers who take the scores on the parts of the rubric and essentially average them into a score for the whole project. I don't do that. You can do very well on a bunch of parts of a project and still not have it gel into a cohesive excellent project. You can have a bunch of mediocre parts and still make something amazing, much greater than the parts. In life, we're graded on overall impression, not by whether we had two misspellings or three.

Thursday, August 6, 2009

Define the problem

The ever-popular Seth Godin has a post about "shaving the bear", which he uses as a memorable analogy for the idea of solving the symptoms of a problem rather than addressing the root cause of the problem. (It's based on a PSA about shaving bears so they wouldn't overheat due to global warming.)

Possibly the best lesson I ever learned from the best manager I ever had was about articulating the problem, not the solution you want. (We'll call the manager TZ.) People tend to ask for the solution they want as though not having that solution is the problem. Sometimes that's true, but more often there are many solutions to the actual problem, and by articulating the true problem, many solutions become possible.

It came up because of this situation: TZ was working an event. A small group approached him and asked if he knew where a janitor was. The answer was no, but TZ stopped them and said, "why do you need a janitor?" Because they were out of toilet paper in the men's room, and a janitor would know where it was stored. TZ didn't know where a janitor could be found, but he turned to a nearby woman and asked if she could go into the women's room and bring a roll or two of toilet paper to one of the men and he could take it into the men's room. Actual problem solved. Requested solution unneeded.

The most memorable time I've applied it was in the process that resulted in my school becoming a 1:1 laptop school. It began with the teachers asking for a computer lab. The thing is, we already had a computer lab, but it was full all the time due to space constraints. Putting in another lab would only provide access to teachers who happened to already teach in that room - the space constraints would still be there. What teachers really wanted was more access to computing, access when they wanted it to enhance their teaching. A lab wasn't the best solution to that problem. A 1:1 laptop program, though oddly controversial, solve that and other problems, like inequitable home access. (Why did I find the controversy odd? We're in silicon valley, we'd be solving problems, and it wasn't going to cost any more than the previous solution.)

The first step of both the engineering design process and the scientific method are about defining the problem. Isn't it fun when life lessons are embedded in science? I wonder if it helps kids internalize lessons when we can show the same idea in many domains.