This year, Edge.com asks of great minds a good question: What is your favorite deep, elegant, or beautiful explanation? And 192 great minds responded. As I read through the answers, I'll comment on those that catch my eye.

Today, I'm going to address the answer titled "You Think, Therefore I Am", from which I'll quote the first few paragraphs (but be sure to go to the Edge.com link above to see this and the rest of the answers):

"I think, therefore I am." Cogito ergo sum. Remember this elegant and deep idea from René Descartes' Principles of Philosophy? The fact that a person is contemplating whether she exists, Descartes argued, is proof that she, indeed, actually does exist. With this single statement, Descartes knit together two central ideas of Western philosophy: 1) thinking is powerful, and 2) individuals play a big role in creating their own I's—that is, their psyches, minds, souls, or selves.

Most of us learn "the cogito" at some point during our formal education. Yet far fewer of us study an equally deep and elegant idea from social psychology: Other people's thinking likewise powerfully shapes the I's that we are. Indeed, in many situations, other people's thinking has a bigger impact on our own thoughts, feelings, and actions than do the thoughts we conjure while philosophizing alone.

In other words, much of the time, "You think, therefore I am." For better and for worse.

An everyday instance of how your thinking affects other people's being is the Pygmalion effect. Psychologists Robert Rosenthal and Lenore Jacobson captured this effect in a classic 1963 study. After giving an IQ test to elementary school students, the researchers told the teachers which students would be "academic spurters" because of their allegedly high IQs. In reality, these students' IQs were no higher than those of the "normal" students. At the end of the school year, the researchers found that the "spurters'" had attained better grades and higher IQs than the "normals." The reason? Teachers had expected more from the spurters, and thus given them more time, attention, and care. And the conclusion? Expect more from students, and get better results.

The authors stress rightly that the Pygmalion effect demonstrates how we can influence another's being with our thoughts. Many believe that the Pygmalion explains what we colloquially refer to as a self-fulfilling prophecy. Think something will succeed and it will be more likely to succeed. Think something will fail, and it will be more likely to fail. This phenomenon appears counter-intuitive. Why should our opinions about how something will result influence the result? If we had just told those teachers that all of the students were promising, one might ask, wouldn't all of the students have done well?

The answer is: maybe. But the Rosenthal study has a more intuitive interpretation in keeping with the assumption that humans are (at least boundedly) rational thinkers (at least sometimes). This explanation, as we'll see, suggests there's no magical stuff going on here. The Pygmalion effect results from teachers acting rationally given the information they obtain about their students's diverse abilities.

It goes like this. Suppose teachers want students to be successful. Suppose also that teachers have limited energy to spend focusing on any particular students. Suppose also that the laws of thermodynamics apply, so that teachers cannot invest energy in two things at the same time. Suppose also that investing energy in a student produces diminishing marginal returns because eventually the student cannot absorb further information. The relationship between the amount a student learns and the energy expended on that student might look something like this:
See how the benefit of investing in a student gets smaller with each additional unit of energy spent on teaching that student? Eventually, the teacher's investment won't matter much at all to how much the student learns. 

Now suppose that we have two students. One is intelligent and motivated: the model student. The other is not so intelligent or motivated. The model student can absorb more information; so the benefit of investing energy in that student doesn't diminish as quickly as for the other student. Suppose that the model student is three times better than the other student. Comparing the learning functions for these students might look like this: 
The blue line is the learning function for the model student. Note how the red line, representing the learning function of the less talented student, flattens out more quickly than for the  model student. This is very important. Here's why.

Suppose the teacher wants to maximize the average success of these two students (actually, you can suppose the teacher wants to maximize the total success of the students; the results won't change). How much more time should the teacher spend on the model student? It turns out that the the teacher should spend on the model student an amount of time triple that of the other student, which is exactly how many times better the model student is than the other student. That's classic expected utility theory for you (some would call it the expected utility hypothesis, because human behavior appears to deviate in some contexts from that predicted by the expected value logic).

Thanks for everything, von Neumann and Morgenstern!

So what does this result mean for the Pygmalion effect? It means that we should have expected the teachers to invest more in what they believe is a better investment. The Pygmalion effect doesn't prove anything about self-fulfilling prophecies or any other such mysticism. It's just basic thermodynamics and classical economics. It also demonstrates a harsh reality of teaching: If teachers aim to maximize the average success of their students, they're going to focus less on the students who need the most help.

(Interestingly, if there weren't diminishing marginal returns to the teacher's investments, you'd expect the teachers to flip the bird to everyone but the best students.)

When it comes to my own research on migrants' decisions about how much money they spend on their family and friends back home versus themselves, diminishing returns is a crucial assumption. Whether migrants remit because they have altruistic inclinations toward their close kin, or they're maintaining relationships for self-interested reasons, they wouldn't send anything if there were linear or increasing returns to the dollars they kept in their own pockets. Since migrants from developing countries send more money to developing countries than official development aid organizations, we should be thankful for the existence of diminishing marginal returns.

For my own part, whatever the reason I care enough about my own family to send something home every month, I'm also thankful for diminishing returns. In fact, if there were no diminishing returns to anything at all, neither of us would have gotten out of bed this morning. Chew on that.
 
 
If you've ever taken a statistics course, you've probably heard of the birthday problem. If you haven't, the birthday problem asks what the chances are that two people in a room full of, say, n people have the same birthday. It's a fun problem to solve in a class of 23 or more people because it turns out that there's a greater than 50% chance that two out of 23 individuals were born on the same day. More importantly, it turns out that nearly 100% of probability theory newbs think the chances are much lower.

Also, I pulled one of those statistics out my butt. Guess which one, then realize it might not be far off the mark and forgive me.

Today, I was reminded of the birthday problem while I was reconciling church birth records with a genealogy for this village, which I've been improving since 2010 by combining key informant interviews with archival research. One of the last records I checked was a woman whose first initial is J., and who was born on my birthday in 1927, 57 years before I was born. 

"Wow," I said to myself, "I found someone with my birthday, what are the o..."

At that moment, the Debbie Downer of statistical reasoning blew her sad trumpet to remind me of the birthday problem, which I remember from one of my game theory courses in which we reviewed probability theory before getting into the Bayesian equilibrium concept. No, Brash, sighed statistical reasoning, it's not that special.

I felt like most women feel when I them that the whole menstrual synchrony thing is (probably) complete bullshit, as Cecil Adams summarizes with his usual aplomb.

I slapped myself out of my statistical-reality-induced funk and thought, "Actually, it's kind of cool that I'm almost guaranteed to share a birthday with one of these people." Then I hopped over to Wikipedia to review how to calculate the probability to avoid thinking hard at midnight after 12 hours of work. Finally, I calculated the vanishingly small probability that I don't share a birthday with any of the 1,326 records in my genealogical database.

I'll skip the details and tell you that the probability is zero. Here is a good example of how scaling up a problem can help you understand it. We can reduce the birthday problem to the absurd and envision the probability that I don't share a birthday with at least one other person on Earth, which currently nourishes about 7 billion sacks of living, breathing, human meat. But you'd think I was an idiot if I thought about calculating the probability in that case. Of course someone shares a birthday with me. Hell, on my birthday this year, about four people will be born every second.

Thank about that. The second you were born, three other beings were likely screaming their way into this confusing, beautiful world. Each birthday you have you share with enough people to enlist in almost 4,000 Roman legions.

I only wish that the scaling up problem worked in my personal life. Alas, I have an n of only three. The simultaneous trials and triumphs of parenthood, grad school, and living apart from my family must remain seeming paradoxes until I get the time to sort them out the old fashioned way.
 
 
The only popular thing I've ever written is a flawed attempt at snarkily summarizing the results of a peer-reviewed journal article about a voluntary public goods game with a destructive opt-out strategy. The article I covered was published in the Journal of Theoretical Biology. It's trending on reddit.com for the second time (thanks for the heads up, B.T.!).

The piece is terrible looking back on it. I was trying my best to appeal to the website's audience (people who like Cracked.com) but also cover an intellectually stimulating topic within my own field. As many in the reddit thread note, it just made me seem like a pompous a-hole.

Then again, that was the purpose of creating my Brash Equilibrium persona in the first place. Brash is a pompous, pretentious, albeit badass man of mystery who sports a mustache, cowboy hat, and a smug grin, and who crosses his legs Euro-style like a freakin' boss.

Oh, dear. That reminds me that I actually wrote another God-awful thing that went viral on the topic of crossing your legs Euro-style.

Ah. Writing snark for extra cash while also trying to be a legitimate social scientist. That was a fun year.
 
 

(Sweet Jesus this post needed some editing!)


One of the missions of my field work in Dominica is to develop a comprehensive genealogy for this rural community so I can measure the relationship between kinship and sharing between migrants and non-migrants.

It's common knowledge that there are a few very common surnames in this village, but the conventional wisdom doesn't cover the rest of much less common surnames. I tabulated the frequency of different surnames and ordered them by rank to learn more.

Below is the discrete distribution of surname frequency by surname rank (higher rank means higher frequency; there were lots of ties, especially at lower ranks). Because there is good reason to believe that the distribution of surname rank should follow a power law distribution, I also fit a power law and calculated the R-squared value.

A power law, by the way, simply means that the frequency of something that interests you (in this case, surname rank) can be calculated using a formula like this:

(frequency of surname rank) = a(rank)k

where "a" and "k" are parameters that you estimate from real data.
From the R-squared (and the graph), it's clear that the power law model does a good job explaining the variation in surname rank, especially for such a small sample (small relative to, for example, the sample of surnames this guy used in his analysis).

Because we'd expect surname distributions to follow a power law, you might think this analysis isn't particularly surprising. And you'd be right. While the ubiquity of the power law in nature and the reasons thereof are fascinating, it is a pretty well studied phenomenon.

What I don't think is well understood is how universal the parameters ("a" and "k") of the power law distribution are in a given context (i.e., in the distribution of surname rank across different populations), and what mechanisms influence variance in those parameters. Similarly, the fit of the power law distribution to actual surname rank data seems to vary somewhat across populations. Why? The answers to both of these questions may have something to do with the cultural, social, demographic characteristics of a population.


In this community, for example, high rates of out-migration (a net loss of about a fifth of the population between 2001 and 2011) likely affect the current distribution of surnames. Members of the same family tend to migrate together, the psychology underlying the motivations to migrate may run in the family, and some families tend to migrate to destinations where their family members already reside because it eases the transition.

How does this population process influence the power law (and fit thereof) of surname rank relative to, for example, a growing population with rapid IN-migration?

I don't know and I don't have much time right now to investigate. But it's an interesting question.
 
 

Did you know that there is a website called Whale.fm where you can do your part to help marine biologists decipher the eerily beautiful language of whales?

There are all sorts of similar research projects out there tapping into the wisdom of the crowd to deal with difficult problems and large datasets. It's been called crowd science or citizen science. Thanks to the Internet, humans are inventing new ways of thinking that aggregate our grey matter.

Someday I'll have a genealogy or some other dataset large enough so that the time investment in developing a citizen science platform would be worthwhile. For now, I'm thinking about embarking on what I will call open source noodling. And I want you to tell me whether or not this is a really bad idea.

Open source noodling puts together two terms, "open source" and "noodling". Open source, of course, refers to a mode of production that involves free access to and distribution of an end product's design. Many argue that open source bolsters innovation and levels the playing field. Scholars in many disciplines, including my own, are debating a shift to open access peer review from the corporate, for-profit peer-reviewed journal model. Open access is very similar to the open source philosophy.

Noodling is a term I first heard in conversation with Richard McElreath, an anthropologist at UC Davis who studies cultural transmission and lots of other stuff. Basically, it's the use of abstract mathematical models to answer concrete questions like, "Why are humans so reliant on social learning?" or..well..anything really, if you build the model properly.

If you put open source and noodling together you get the development of a mathematical model to answer a question, which is built collaboratively using a system that allows open distribution and free access to information.

What I want to do is take a mathematical model I've been working on, and which is basically finished, write a draft manuscript, and post it in a medium that allows for open discussion and possibly even open editing. This sounds like a terrible idea but I'm playing with it, and want you to come and play with it, too.

The benefits of open source are manifold. I could tap into a huge knowledge base to solve a problem that I think is really interesting. It bypasses the slow process of peer review while also potentially improving a manuscript in preparation for the peer review format. Yet open source also poses new problems about who gets credit for the research, how to filter information, and how to aggregate the wisdom of the crowd into a singular product.

How would you set something like this up? I encourage open (heh) discussion about it. Or contact me personally, whatever. It's an idea I'd like to develop into a systematic method of producing and distribution knowledge.

I'd also appreciate it if you told me if I was reinventing the wheel, and if there are already venues out there where open source noodling can be done in a way that maximizes the benefits but minimizes the costs.

Anyway, here's an example of something I'd like to submit to open source noodling. I have other things, too, but here's one for starters

In between genealogical interviews with key informants and preparation for the rest of my field surveys, I'm working on some side projects to pass the down time.

The one I'm focusing on now is part of a project I'm calling "the tragedy of hawkish altruism". The tragedy of hawkish altruism is this:

If individuals who are aggressive to members of their own group make that group better prepared for competition with other groups, it may drive the evolution of aggressive behavior that leads to more conflict within groups. In this scenario, hawkish bastards are also tragic heroes because they risk their necks training for and maneuvering on the evolutionary battlefield against other groups. 

Think about warriors risking severe injury to realistically train for battle. Think about office intrigue and backstabbing that prepares people for the rigors of negotiation with other firms.

The tragedy of hawkish altruism ties into a broader debate in evolutionary biology about the strength of natural selection among groups versus individuals. Natural selection needs variation in order to happen. When individuals move freely between groups, it causes all groups to be more similar to one another and for most variation to exist among individuals. For this and other reasons, evolutionary biologists have traditionally considered individual selection to be stronger than group selection.

Yet a few evolutionary biologists argue that group selection may still be an important force in some cases. For example, the tendency of humans to conform to the customs most common in their current group may allow variation among groups to increase relative to variation among individuals, causing group selection to overpower individual selection. 

Researchers have used this reasoning to support a hypothesis for why humans are an especially cooperative species, even in large groups of unrelated individuals, and even when cooperation requires altruism, which would be selected against among individuals but favored among groups (that characterization of humanity, by the way, is itself a hypothesis that I don't think has been adequately tested; at least not the part where cooperation is assumed to be altruistic). 

Anyway, this so-called cultural group selection explanation for the cooperativeness of humans is contentious for many but compelling for some. Unfortunately, some people I've spoken to during academic conferences and after-school happy hours have clearly misinterpreted the argument as meaning that group selection leads inexorably to socially desirable outcomes.

Um. No. In fact, some research suggests (though not without substantial controversy) that group selection could potentially explain warfare if warfare entails individually costly contributions toward common defense or offensive raids that benefit a group. Hard to argue that warfare between groups is socially desirable.

Back to the tragedy of hawkish altruism. I've developed a mathematical model showing that group selection cannot only explain conflict BETWEEN groups. It might also explain conflict WITHIN them.

Where the crowd science and open source noodling comes in is that I want to post an early version of my manuscript and hold an open discussion about it before I submit it for publication. Hell, maybe the discussion might result in a collaboration. Scientists share their unpublished manuscripts all the time with their colleagues. And there are many many many scientists out there who could be my virtual colleagues. Why limit it to the people I know personally?

My ultimate goal with this tragedy of hawkish altruism stuff is to develop several quantitative models of increasing complexity while amassing a large, cross-cultural dataset to test the assumptions and qualitative predictions of those models. In fact, one of my post-PhD plans is to apply for a Harry Guggenheim Research Grant (they fund research on dominance and violence) to do this research.

Anyway, I'm looking for suggestions on how and if to do this kind of open source noodling in a way that maximizes the coolness and soundness of the science while minimizing the possibility for unforeseen bad stuff to happen. So how about it?

 
 
Our ancestors once common, as are we,
All slaves to wanton anisogamy.
Theirs was visceral; ours is digital,
With zeal suboptimal 'til I return.
 
 
Like most academics, I overstretch myself with side projects and collaborations. Whatever. Now I am in the field, and there is ample downtime despite all the surveys I've got to do.

So here is a list of side projects--two lone wolf, two collaborative--that I hope to finish during downtime.

1. the tragedy of hawkish altruism

I've developed a mathematical model to analyze the conditions when group selection can increase within-group aggression beyond an individually optimal level toward a higher group level optimum when the byproduct of within-group aggression is better performance in inter-group contests. The model is cool for two reasons. First, it flies in the face of a growing fallacy--fueled by mathematical proofs that group selection can in some cases lead to within-group altruism--that group selection is something that leads to socially desirable outcomes. Jung Kyoo Choi and Sam Bowles and others have already argued that group selection can promote inter-group warfare. I'll argue that group selection can promote people being shitty to their brothers in arms.

2. the joker effect?

A while ago, some people published an article in the Journal of Theoretical Biology showing that the presence of destructive individuals can promote cooperation in a voluntary public goods game. I submitted a short letter outlining some reasons why the published model doesn't present a strong solution to the collective action problem. The reviewers didn't like the letter, to say the least, and I cried into my academic pillow before resolving that I'm still worth a damn. I'm still convinced I am right and am going to resubmit the letter as a full article, addressing all of the reviewers concerns and improving the paper greatly.

3. human-environment interaction

I've been collaborating with a sociologist, a public affairs researcher, and a geographer. We are comparing the explanatory power of different measures of environmental quality when predicting rural-to-urban and urban-to-rural migration in Nang Rong, Thailand. We're using a pretty unique, very large, truly longitudinal migration history dataset, and developing some sophisticated statistical tools to tackle the problem. 

My colleagues will present the first paper associated with this project at the Population Association of America meetings this year. This project has a lot of potential. We're assessing the predictive value of easy-to-obtain, global scale environmental measures (El Niño/La Niña-Southern Oscillation events), versus local scale remote sensing data measuring plant health from light reflectance, which requires several steps of time-expensive processing. 

We're using high-level signal processing methods to find environmental measures that match the rules of thumb people use when making difficult decision about where they should live. Anyway, I want to help my colleagues finish this paper and others we've got in the works, to the best of abilities.

4. Parental Investment Vignette Project

Dr. Geoff Kushnik has designed a clever experiment meant to highlight the factors that influence mothers' decisions about the tradeoffs they face when investing in their children. He wants to make this project cross-cultural. And I want to help him do that. Plus it will be pretty easy to get the n of 40 women necessary to participate.

other stuff...

In addition to the projects above, there are three other things I'd like to get off the ground while I'm here, and they're all related. First, I want to assess the need and desire in these communities for community designed and constructed playgrounds using the KaBOOM! community organization methods, but building with local materials.

Second, I've gotten people here interested in the idea of organizing an annual 12 kilometer run/walk event as a way to raise funds for community projects, tapping local business and athletic fervor, plus the Dominican Diaspora. I want to make sure that this is a feasible thing to do, and that people want to do it, and that people want playgrounds before I go too much further.

Third, if the playground thing flies, I want to take the advice of Rick Kauffmann, student of David Sloan Wilson and Evolution Institute affiliate, who said I should investigate the effects of different playground designs on community level outcomes.