October 1, 2012: Faces

Faces is a machine-learning driven art project, inspired by a hallucination, producing images from algorithms similar to those our cameras use to detect faces.


When I stumbled on this drawing of a hallucination, I was struck by how it had distilled the human face into light and dark. My next thought was, “I bet a camera’s face finding algorithm would go crazy on this.” From that, I wondered if I could create similar images using such algorithms.


Cameras use computationally cheap (fast) face detection. Considering sub-regions of the image at a time, they ask questions like, “Is the mean luminance in A greater than that in B?” A and B, and many other pairs of rectangles like them, have been chosen by engineers such that the question will generally be true when a face is present in the image.

The first task was to find a set of rectangle pairs, like A and B, which generally predicted the presence of a face. I preprocessed several images of friends, and generated random rectangle pairs in Matlab. If the mean luminance inside the first was greater than that inside the second for most of my pictures, then I kept that pair, along with a measure of its performance.

To create the new images, I sampled from this set of successful rectangle pairs. I started with a new, black image: where a rectangle pair preferred light, I lightened the image; where it preferred dark, I darkened.

Some of the results were very “facey”, some were not:

I painted two of these faces onto canvas with the help of a projector, but the photos I have of this are horrible.


Programmers designing computer-perception attempt to reduce entities to their most defining characteristics. This helps their programs efficiently identify those entities. These same principles make for good brain design, and so we sometimes find that our most efficient machine-learning algorithms have some kind of implementation in the brain.

This project explores some overlap between defining characteristics and minimal representations in the biological and mechanical domains.

You can find the drawing that inspired me here, and you can read more about the algorithms your camera uses to detect faces here.

July 10, 2012: Staff Cultural Orientation

I’m returned from ten days in the Yukon-Kuskokwim Delta, where I’ll be teaching, having attended a staff cultural orientation with fifteen new teachers and one new principal.

Outskirts of Tuluksak
Outskirts of Tuluksak

We spent five days in Akiachak where the district headquarters is located, addressing the same questions I’ve been asking myself about the imposition of western culture on native peoples. We explored two threads:

  • The Alaskan sub-arctic is a dangerous and difficult place to live. Yup’ik culture, a product of thousands of years of social evolution, contains a set of implicit rules for thriving in this environment. The language encodes a mindset or worldview that helps its speakers adapt to living here (see the Sapir-Whorf Hypothesis), and this effect can’t be translated into and transmitted through English. By interrupting the generational transmission of native culture, we are interrupting the transmission of this worldview and set of rules which allow the people to survive the sub-arctic.
  • Through the media and education, western society imposes a worldview and a set of values on the native people. For a variety of geographic, economic, and social reasons, most Yup’ik are unable or do not want to participate in the mainstream western lifestyle. This creates a psychological conflict, which may result in questions about identity and feelings of inadequecy among people who have been made to feel that mainstream western society is superior to their own traditional society. Many native populations are in absolute psychic crisis because of this conflict.

The message was: accept your own ignorance and seek to be an aid to the residents, not a guide.

The other five days we spent in our own villages. We were assigned to wonderful mentors among the village residents, who invited us into their homes and introduced us to a bit of their subsistence lifestyle. We also had a chance to meet those of our students who were attending summer school. The kids definitely try to see whether they can get under your skin, but I feel I passed their tests.

I’ve started to read Enlightenment’s Wake by John Gray. Gray writes that the fulfillment of the Enlightenment is the central project of western civilization. In my own words, the Enlightenment is the instantiation—the particular form, realization, or instance—of the Conqueror Mindset in western civilization; other cultures would have different projects of progress centered on the universalization of Sharia and Islam, Confucianism, etc.

So as westerners, we believe that science and reason can help us produce a template for a prosperous society with a minimal set of rules that allows people to fulfill their own independent, personal values. Since the template allows people to define their own personal values, it can in theory be lain-over any population without disturbing local culture, so long as the population chooses to value and retain that culture. But Gray points out that independence is itself a value that is taken for granted by this Enlightenment project, and that we may be better served by a model of civilization which does not allow for independent assignment of value.

By my interpretation, if you lay the Enlightenment project over a culture which assigns value collectively then the native culture will crumble and fail. More thoughts to follow.

April 21, 2012: The Conqueror Mindset

I now have a contract, for teaching secondary mathematics in Tuluksak, AK:

View Larger Map

Tuluksak is populated almost entirely by native Yup’ik. Anyone can guess that the culture will be different. But how different can two human cultures really be? This videogot me thinking that the answer could be “very different”. The relevant quote:

And performance is the bottom line in the culture of the lower 48. Here [in rural Alaska] it’s not performance; it’s family and community. So there’s a real conflict there, between the two ways of life. I had to give up the performance approach to the community/family approach.

–Richard Stasenko

Why the drive to perform? Almost every major culture on earth is descended from a conquering society. My ancestors were bumming around western Europe until they were conquered by either the Romans or the Vikings. Latin America was conquered by the people conquered by the Romans. What started as a small nucleus of kingdoms in China eventually made most of East Asia into vassal states. India, North Africa and the Middle East, Sub-Saharan Africa, and the pre-Columbian Americas have similar stories.

The majority of human cultures, including the native cultures of your ancestors, have been displaced by a small handful of empire-building cultures. Each of these few cultures ascended because they cared about performance; if they hadn’t cared about performance they would have been absorbed by some other, more effective culture. We are the cultural descendents of performers. They have passed to us the conqueror mindset: a desire to perform beyond the needs of survival and comfort.

Ran Prieur likes to point out that there are cultures which have no concept of freeloading. If we enjoy our labor and we’re able to survive comfortably, why should we care whether others benefit enough from our work that they themselves don’t have to work? Ran suggests that it’s because we don’t actually enjoy our work. I’ll also blame the conqueror mindset. The conqueror mindset says that there is always work to be done towards the advancement of society, that we must continually progress in our capabilities, and that everyone can and should contribute to that progress.

The conqueror mindset is the engine of science and industry which continually asks, “How can this be done better?” The conqueror mindset pursues “better”, “improvement”, and “more”. Those are constantly rising bars, standards which can never be reached. A culture without the conqueror mindset, though, might pursue “enough”, or “comfort”. These are goals that can be reached, goals that you can reach and then stop working towards for awhile.

Consider that it took us ten-thousand years to go from pottery to metalworking. And the atlatl may have been the height of technology developed in the thirty-thousand years of fully-modern humans preceding pottery. These transitions would not have taken millenia if “better”, “improvement”, and “more” were fundamental human drives.

Until the mid-20th century, the Alaskan interior didn’t seem to have many resources worth conquering. Like with many native populations, there was continuous and low-level inter-tribal warfare until western intervention. But the region never produced or attracted the serious attention of a real empire-building culture until America found the whole region well placed for a defense against the Japanese, and saturated in oil besides.

When Richard Stasenko says that the Yup’ik he taught are not focused on performance, it says to me that their culture was not produced by conquest. And it has some really difficult implications for my role as a teacher. Western education does not teach you how to survive comfortably, to enjoy life in your community among your family, to know when you’ve done enough. Western education prepares for the jobs of tomorrow, prepares for excellence, prepares you to extend human knowledge, to be a contributing citizen. These are the goals that were given to us by our conquerors. What do I teach to a person who has no interest in the jobs of tomorrow?

It’s not a hopeless question, just my own prompt for thought.

April 20, 2012: The Serenader

I’ve got to break up two heavy posts, here, with some more accordion.

Your browser does not support the audio element.

Apologies for my mic topping out. I’ll turn down the boost next time.

April 17, 2012: Praxis II Mathematics, 0061

Studying for this month’s Praxis II Mathematics Content Knowledge test was tough. Every source on the internet claims that there are no good professionally produced practice tests; people find the real thing terribly more difficult. So I wasn’t comforted by the fact that I was doing well on all of my pre-tests.

I think they must have eased up this season, because the test was only somewhat more difficult than the ETS-produced practice test. Based on my performance in practice, I’ll guess I missed 3 to 5. If you’re concerned like I was, you can benchmark your progress against my history:

  • I finished the Cliff’s Notes (2006) tests in 60-75 minutes each, scoring 40-43 per test.
  • I reviewed each of these tests, and recompleted them in 45-60 minutes per test.
  • I did this guy’s test. It took me about an hour (it’s a half test, so that’s equivalent to two hours on a full test) and I got 21 (or 42 on a full test).
  • I completed the ETS-produced practice test in about 75 minutes, scoring 43.

When test day came I spent 90 minutes on my first pass, marking maybe 8 questions as educated guesses. On my second pass I was able to turn 7 of those into solid answers.

I’ll post an update when I get my score back!

I hope that helps anyone who’s freaking out about the test based on message-board chat. The Praxis2Math.com test was definitely tougher than the actual test I took. The ETS produced practice test was only somewhat less difficult than the actual test, and is worth the $14. The Cliff’s Notes (2006) tests were good practice, but definitely much easier than the actual test. The Cliff’s Notes book was a good book, though, to review principles from.

Update: I scored a 181

February 24, 2012: Eine kleine Akkordeonmusik

It’s not spectacular, but I wanted to share my progress so far on the accordion. It’ll be a nice benchmark to look back on in a couple of months. I’m taking lessons from Duane Schnur’s Accordion Site.

Is the song about an insensitive monkey or a hypersensitive weasel? Both!? Neither?!?

February 18, 2012: A Mathematician’s Defense of Functionalism

Mind and Function

A central question in philosophy of mind is what constitutes the basis for conscious experience. One branch of modern philosophy—functionalism—claims that any system which functionally replicates human thought and behavior would also replicate human consciousness. Other philosophers suggest that functional equivalence to the human mind is insufficient for producing consciousness, that conscious experience transcends the ability to produce thought or behavior.

I wrote this essay just as I was moving from advanced math into brain and behavior. It’s a mathematician’s defense of functionalism against the arguments of John Searle, the Chinese Room thought experiment, et al.:

Functionalism, the idea that mental states are characterized by their function, formally arose in the mid 20th century in response to earlier attempts at demystifying the mind(Levin, 2009). Many felt that behaviorism had failed to describe the mind at an appropriate conceptual level, or that attempts such as identity theory were overly restrictive upon the conditions of mental states(Levin, 2009). Functionalists declared that the particular physical processes underlying mental activity in an entity were inconsequential; what mattered was the functional role a state played in that entity’s existence.

…In this paper, I don’t try to prove that functionalism is true. Opponents of functionalism will always be able to find harbor in the the ineffable, the empirically unobservable. I only wish to show that functionalism is valid, that it’s plausible, that it’s potentially consistent with our experience of the world.

…We can formalize this notion a bit more by appealing to the mathematical formulation of isomorphism. Let G and H be sets, and the function f be a one-to-one mapping between them. Let ★ be the function from G to G that we wish to preserve. Then f is an isomorphism with respect to ★ iff f[★(a)]=★'[f(a)] for all a in G<read entire paper>

For some good introductory point/counterpoint on the topic of functionalism, read John Searle’s Mind: A brief introduction and Daniel Dennett’s Conciousness Explained.

November 29, 2011: Projects Added


A machine-learning driven art project, inspired by a hallucination, producing images from algorithms similar to those our cameras use to detect faces…Read More

A Python MUD

An as-yet untitled Python MUD base atop the Miniboa telnet server library…Read More


A toolkit for running psycho-visual experiments in Matlab…Read More

A Modern Trivium

Modern education does not teach us how to think, and only barely teaches us how to learn. In contrast, Medieval formal education began with these very topics, taught via the Trivium. I’m interested in combining the Trivium with models of self and group driven education…Read More