What’s cool is getting data and hacking yourself. What happens in your blood when you eat a low-carb meal, for instance? Now I know, precisely. (Blood glucose actually _down_ from fasting levels — 85 to 76, this morning.) There’s a hypothesis that under fasting conditions if you impose a metabolic demand that glucagon and adrenaline will shoot up to make up the shortfall. Am interested to see how that plays out.

The ultimate point is to investigate how blood glucose tracks with energy levels, pain, mood. I will know eventually. Also will get some Ketostix, since I have anecdotal evidence on extended low-carb’s efficacy for pain control.

I wish there was some more shit I could measure. This was one of the big reasons I applied to both the neuroscience and psychology programs, btw — easy access to do my own blood work. Of course, I didn’t mention that in the application.

Maybe I should get a job as a lab tech after this? Hmm. Or better yet: can you buy the equipment they use to do a full blood workup? I’m sure it’s crazy expensive, but maybe after the initial layout running a single test isn’t too prohibitive?

I got it. Five hundred thousand dollars, bitches.

Well, okay. Twenty-one thousand dollars. Even so.

For instance, the other day I recalled a famous passage from Adam Smith that I wanted to cite: something about an earthquake in China. I briefly considered scouring my shelves in search of my copy of The Wealth of Nations. But I have thousands of books spread throughout my house, and they are badly organized. I recently spent an hour looking for a title, and then another skimming its text, only to discover that it wasn’t the book I had wanted in the first place. And so it would have proved in the present case: for the passage I dimly remembered from Smith is to be found in The Theory of Moral Sentiments. Why not just type the words “adam smith china earthquake” into Google? Mission accomplished.

Of course, more or less everyone has come to depend on the Internet in this way. Increasingly, however, I rely on Google to recall my own thoughts. Being lazy, I am prone to cannibalizing my work: something said in a lecture will get plowed into an op-ed; the op-ed will later be absorbed into a book; snippets from the book may get spoken in another lecture. This process will occasionally leave me wondering just how and where and to what shameful extent I have plagiarized myself. Once again, the gates of memory swing not from my own medial temporal lobes but from a computer cluster far away, presumably where the rent is lower.

When I was in the AI lab at USC we always talked about building an AI, teaching it in various clever ways — teach it like you teach a child! — and this is a clever way of thinking about AI that various AI labs are starting to catch on to, although the main thing about AI labs is that they don’t really like to be called AI labs anymore since the term is so effusive, and they know it.

But we don’t spend enough time thinking about cognition as an enterprise, something to be done cooperatively, even though we do it all the time: Monica knows where everything is in the house; I know the best way to make a pizza. She puts it in the oven, I check it, or tell her when to check it. She tells me where to find my socks. This is a stupid example but it’s one everybody can understand.

But there’s another level, one we’re just starting to see plainly because it’s so plain, and that level is using tools to think, to remember. And most of what’s happened so far has been an accident, as in the above quote, and the tools help us with our “thinking” almost as an afterthought. I’ve been interested for years in making a tool that helped us think better, whose main purpose would be for that and not anything else, but various things have kept me from it.

Someday, when Wes graduates, we can start building it.

And this is why we shouldn’t be surprised that a chess prodigy raised on chess computer programs would be even more intuitive than traditional grandmasters. The software allows him to play more chess, which allows him to make more mistakes, which allows him to accumulate experience at a prodigious pace.

So what’s the lesson, regarding whatever thing you do, or whatever you want to do? Whatever is important to you?

Do a lot of finished work. Don’t worry so much about great work, brilliant work, groundbreaking work. Finish stuff. The end.

Here it is, for your edification. An interesting nugget is that this handful of words took a god-awful amount of time to write. I could have written five stories in the same amount of time. Hopefully that will get easier with experience.

*

Overview

Why do we like the things we like? We prefer certain paintings to others; music and literary critics expound at length on what makes work X superior to work Y, often against popular opinion; and all of us have had the pleasant frisson of finally figuring out where the oddly-shaped puzzle piece fits. We have strong affective feelings about a wide range of stimuli, for no obvious reason. Why is this?

It’s not surprising that people have been trying to formulate a rationale for aesthetic preferences since Aristotle. Sometimes they agree on general principles, often they do not; and the candidate explanations are either topic-specific or else too vague to operationalize. But what if a single quantifiable mechanism undergirded these disparate pleasurable phenomena, from simple aesthetic valuation (like appreciating a Magritte painting) to more abstract pleasure (like finally figuring out a calculus problem?)

They say there’s no accounting for taste, but I think they’re wrong.

Research goals

The discovery of mu-opioid receptors in the parahippocampus (a principal associational ‘mixing area’ at the end of the visual pipeline), and the research that has followed that discovery, motivates a powerful idea: we get a particular chemical reward when we find patterns in the world that hit a sweet spot between predictable and chaotic. The ‘sweet spot’ comes from parahippocampal neurons that have, via competitive learning, come to code for specific patterns of visual stimuli, but whose codes have not become so ‘overfit’ through experience that spurious activity has been eliminated. Stimuli that are ‘understood’ by the population code result in an opioid rush proportional to the neural engagement, even as environmental regularities rev-down the neural (and opioid) response.

To this point the idea has been developed perceptually; but neuroanatomical and clinical evidence suggests that, far from being particular to late visual cortex, this ‘drive-to-pattern’ is fully general, and that it transcends perception and intrudes into more abstract (and purely cognitive) domains. If this is true, then we can adapt statistical coding models from perception, combine it with the mathematical formulations of information theory and knowledge-representation theories from cognitive science, and test whether generalized cognitive pleasure (CP) is subject to the same principles that characterize affect from visual scenes.

Research Plan

There is an extensive memory literature on serial and sequence learning that could be fit to the CP paradigm. The idea is this: a subject is presented with a stream of symbols, generated according to a context-free grammar. The subject’s task would be to discover ‘legal’ patterns from within the symbol stream, collapsing it as the patterns are induced. Each successful ‘collapse’ should reward the subject above what would happen in a control task featuring randomly-generated symbols; we predict that subjects would voluntarily spend more time on the task with symbol streams generated by inductable grammars. Further, we might expect differential results from subjects given ‘hints’ about the top-level structure of the grammar, vs. those given hints about bottom-up structure; and a host of variations on this theme.

To go beyond the rather simple sequence task described above, we need a way to generate a test corpus whose statistical characteristics can be altered to reflect patterns of differing complexity, and which can be extended into domains more sophisticated than streams of amodal symbols. Further, we need a way to quantify collected data in terms of the latent structure inherent in it. Fortunately, I have a head start on this task: last semester I implemented a machine learning system for hierarchical topic modeling using Latent Dirichlet Allocation. Adapting this LDA solution for the CP domain would allow us to generate pattern-laden stimuli in a host of modalities, from visual patterns, to musical compositions, and even text segments. We could use these stimuli as source data for the sorts of experiments described above, as well as for more ambitious experiments.

My advisors’ unique strengths will be put to use throughout this process of theoretical formulation and experimental design. For instance, the memory of chess masters for legal vs. illegal chess board configurations demonstrates that pattern recognition and memory are inextricably linked; and the CP model, which is deeply tied to pattern recognition, has much to gain from these results. Wilma Koutstaal’s encyclopedic knowledge of the memory and learning domain has already proven an invaluable resource in my literature review, and her suggestions of which results might be re-purposed to address the question of CP have been a great help even at this early stage. Further, our work on my first year project on the neural correlates of abstract vs. concrete semantics suggests several obvious next steps once the preliminary proof-of-concept work already discussed is finished.

Paul Schrater’s expertise in pattern recognition and mathematical modeling finds natural expression in the information-theoretic formulation of CP, and his current work on ‘aspiration’ uses some of the same statistical modeling techniques I propose to use for data generation and analysis. Much of Paul’s work involves how perception of regularity in the environment alters behavior, and insight from those results will inform my experimental design from a computational standpoint much as Wilma’s insight will inform it from a memory perspective. Further, Paul’s knowledge of both computer and biological vision will be critical in extending the symbolic presentation protocols of the first experiments into the image domain.

My wish-list for experiments past those described here are too numerous to list. The chief power of the CP idea is that results from disparate areas that can be stated in information-theoretic terms are suddenly amenable to analysis. Which means CP could bridge, experimentally, theories of higher-order cognition, perception, and aesthetics that have thus far been considered distinct islands of inquiry.

If your girlfriend lights your hair on fire and the police ask you what happened and you launch into a chemistry lesson on oxidation then you are highly educated but you are also a dumbass.

Hopefully you feel enlightened now.

In human development, primatology, and artificial intelligence, the notion of contingency matters way more than you’d imagine. Contingency, in a phrase: I do something and it causes you to do something. Contingency matters in human development because infants react contingently to certain stimuli, which allows developmental psychologists to intuit various things about their cognitive abilities; and because, like with Arthur, the infant’s contingent reaction to its mother is an important part of the bonding experience. Later it will matter in a host of shared-attention features necessary for language use; these kind of contingent reactions is the most important thing that separates humans from non-human primates (cognitively speaking; other differences matter more when it comes to dating, for example.) Contingency’s role in AI is discernable from those already mentioned: devices that react to you seem intelligent — you call, the robot comes running. The effect works like magic.

What I find remarkable is how powerful this simple idea is, even in day to day life. I like the waitress who says something in response to the stupid joke that I said to her, something that indicates that she has actually heard the joke. I do not like the snotty lesbians at The Spy House because nothing I could do would seem to alter their condescending attitudes. Come in with my pants around my ankles? Yawn. Come in with Arthur in a backpack, wearing a little Vikings helmet? Just more tiresome visual stimuli their retinas are forced to transduce. You’d think they’d grown up in that bar from Star Wars where Han Solo shoots Greedo.

Anyway. Here’s a tip, for free: to be twice as charming as you already are, behave contingently. Listen to what people say to you, and then say something back to them that contains the seed of what they’ve said. There’s interesting work to be done here. You can get a long way in life without mastering much more than this one skill.

But now I wonder if it’s a skill that can be mastered, if you’re not predisposed to master it. If chimps can’t learn to do declarative pointing, can people learn to listen, to _really_ listen? Hmm.

If you were online in 1996, you remember some of the anything-is-possible topsy-turvy feelings. But more important, I think, is the history of thought that accompanies any potent tech advance; where zealous practitioners say: this changes everything. Like this, at the dawn of radio:

Ditto, concurred a colleague: “Let a legislator now commit himself to some policy that is obviously senseless, and the editorial writers must first proclaim his imbecility to the community. But let the radiophone in the legislative halls of the future flash his absurdities into space and the whole state hears them at once.”

Obviously it didn’t work out that way; and it would be easy to get cynical, and say that nothing is ever any different, which is the same kind of stupid dichotomous thinking that causes a lot of the problems that surround us every day. A rule of thumb is that no technology is likely to change the essence of human nature; but human nature can be, and is, tweaked all the time. So it’s interesting to see how many times these utopian prognostications have been made, and what they’ve been made about. And what has really been changed, and what hasn’t.

The response to Ardi’s unearthing was not surprising. According to surveys, barely a third of Egyptian adults have ever heard of Charles Darwin and just 8% think there is any evidence to back his famous theory. Teachers, who might be expected to know better, seem equally sceptical. In a survey of nine Egyptian state schools, where Darwin’s ideas do form part of the curriculum for 15-year-olds, not one of more than 30 science teachers interviewed believed them to be true. At a private university in the United Arab Emirates, only 15% of the faculty thought there was good evidence to support evolution.

From this article in the Economist.

I think stuff like this is good for me to see because it reminds me that even though there a lot of dumbasses in this country – A LOT – there are even more dumbasses elsewhere. And in fact, our homegrown dumbasses are like fucking Rhodes Scholars compared to your average global citizen.

This is something very very easy to forget, and while at first I found it kind of hopeful – perhaps we’re not as bad as I thought – upon further reflection it is not at all hopeful. Because it’s not a zero sum game, and having it made clear that Egyptians and vast swaths of the Middle East are ignorant out of the stone ages does not make our own position as Americans any stronger. It just makes the whole world worse.

Bleh.

When it’s been inconvenient for me to believe this I’ve made all kinds of justifications about why it isn’t true. But the rest of the time, I’m fully on-board with the idea that your social groups will determine who you are. It would be a fun experiment to go all the way with this – imagine that, instead of introducing you to my friend Swanky, I just took you around to all the groups of which he is a part. Took you to where he works, the bars he drinks in, to meet the people he hangs out with. I have no doubt you’d have a pretty decent idea of what he’s about, without ever meeting him or asking anyone anything about him.

Swanky would be an extreme example because he lives such a broad and socially connected life; but I can even imagine this being true for other people, who instead of having, say, twenty principal social components to their social groupings, have only two or three. In that case, the theory goes, they would be far more heavily defined by each particular social grouping, since there’s so few of them. Which would make each one really important. If you only have two friends, you’re going to be defined by those two people. Scary. Or maybe wonderful, depending on the people.

Anyway, I’d read journal abstracts of this research with regard to physical components like obesity and longevity, but now it’s in the popular press:

The data exposed not only the contagious nature of obesity but the power of social networks to influence individual behavior. This effect extends over great distances—a fact revealed by tracking original subjects who moved away from Framingham. “Your friends who live far away have just as big an impact on your behavior as friends who live next door,” Fowler says. “Think about it this way: Even if you see a friend only once a year, that friend will still change your sense of what’s appropriate. And that new norm will influence what you do.” An obese sibling hundreds of miles away can cause us to eat more. The individual is a romantic myth; indeed, no man is an island.

From Wired.

When you think about this stuff it makes you swallow hard and ask yourself: are the people I surround myself with the kind of people I want to become? Because you will. You are. We all are, turning into the people we spend our time with. Hopefully it’s a transformation to be proud of.