Sunday, March 28, 2010

Save The Social Animals

Michael Vassar has a fascinating comment on Less Wrong:
[M]y guess is that suffering is an evolved elaboration on pain unique to social mammals or possibly shared by social organisms of all sorts. It seems likely to me to basically mediate an exchange of long-term status for help from group members now.
People often complain that it is difficult to construct an algorithm that can choose which animals we should favor and which we should not, given our limited resources (for example, see Jonah Lehrer here). Too often we rely on the cuteness factor. If correct, Vassar's theory has clear policy implications if we want to reduce animal suffering: save the social animals.

####

This essay argues persuasively that the most utilitarian suffering to curb is not that felt by animals in captivity, nor by other humans, but instead by animals that live in nature

Monday, March 22, 2010

Violent Video Games And Youth Violence

Tom Jacobs concludes his recent account of this subject by asking whether our "instinctive antipathy toward censorship of any sort [is] blinding us to a growing problem". First, I don't think we have an instinctive antipathy towards censorship. Historically, censorship has most often been used as a tool for certain groups to gain power or status over other groups, and for that reason cultivating a wariness towards it is justified. Second, I don't think that violence in video games is a growing problem, based on two key pieces of evidence:

1) Total US violent crime per capita has been falling steadily from 1993 until the BJS's most recent data point in 2007 (from here):

Moreover, Harry Jenkins notes that the "rate of juvenile violent crime in the United States is at a 30-year low." So, at least in the US, it's not like we need to actively search for something to explain why violence might be on the rise.

2) The most recent meta analysis (abstract here) finds that the longitudinal causal effect of playing violent video games after adjusting for initial aggression and sex has an r value of 0.152, for a percentage variance explained of only 2.31%. The authors of this study argue that small effect sizes can have high practical significance if they accumulate over time or if high proportions of the population are exposed. That may be; it's hard to say. But we do know this: 2.31% variance explained is not a particularly large effect size.

The relevant question should not in this case (nor ever) be reduced to whether a statistically significant effect exists, "yes" or "no." Instead, "how much" should always matter. The poison is in the dose.

(Thanks to Robert Wiblin for sharing the Jacobs article, and to Vaughan Bell for being an inspiration to bloggers everywhere.)

Sunday, March 21, 2010

Depression Hypothesis Responses

Jonah Lehrer's widely read article on the "upside" of depressive episodes has sparked a lively debate. David Dobbs offers good commentary with which I mostly concur:
The sensitivity or orchid hypothesis asserts that these short SERT variants don't make people more sensitive to bad experience but to all experience, bad or good. It's not a "depression gene" but a sensitivity gene.

What you make of that sensitivity naturally depends on other assets or experiences you have. If this is so, then it's possible that this sensitivity hypothesis may account (wholly or in part) for findings that "depressed people" have more insights or creativity -- only it's not necessarily the depression that generates the insight, it's the heightened sensitivity.in
This makes more sense to me from an evolutionary standpoint. There is a spectrum of sensitivity to depressive tendencies, probably mediated by a fair number of distinct genetic polymorphisms. An individual that falls far to the depressive end of this spectrum may be more susceptible to depressive episodes, depending of course upon his "other assets or experiences." But this individual will also be more sensitive to other stimuli, which presumably will be advantageous to him in making babies or in making his babies more likely to survive to make more babies. So, depressive episodes could be correlated with creative impulses without the causal relationship that Lehrer talks about.

In the comments of Dobb's response the Neurocritic adds,
You raised valid points about the complexity of some of the issues, such as clinical diagnosis. But after examining the Andrews & Thomson Psych Review article again, it seems like an even bigger crock of sh*t. ART is cavalier about any distinction between sadness and severe depression.
Indeed.

Saturday, March 20, 2010

Book Notes Investigative Pathways

In my effort to understand what motivates and differentiates successful scientists, I've been trying to read up on of the history of science. Investigative Pathways by Frederic L Holmes tries to draw general conclusions from looking at the day-to-day activities of six scientists, focusing especially on Antoine Lavoisier's work on oxygen and Hans Adolf Krebs's work on the citric acid cycle. Much of it is written in academic and somewhat stilted prose, and he does not shy from describing minute detail of the scientist's work, but he does end up drawing some strong generalizable insights. Below are what I consider the best quotes from his book:

"Krebs learned from Warburg [his mentor] the precept that one should not be afraid to "attack the great unsolved problems of his time," and to find the solution by doing many experiments without hesitating over whether one experiment was worth doing."

It has not been shown statistically that the "most effective way to win a Nobel prize is to be trained by a Nobel prize winner."

"Above all, what [teachers of special distinction] teach is a high standard of research. We measure everything, including ourselves, by comparison: and in the absence of somebody with outstanding ability, there is a risk that we easily come to believe that we are excellent and much better than the next man. Mediocre people may appear big to themselves (and to others) if they are surrounded by small circumstances. By the same token, big people feel dwarfed in the company of giants, and this is a most useful feeling."

"The very high degree of specialization within science that can, on the one hand, demand prolonged periods of training to master its deeper conceptual structures and complex investigative methods, may also open niches in which the young investigator can more rapidly reach the forefront." (as opposed to artists)

"Are the forms of insider access, some of which each of our subjects enjoyed, so essential to a successful investigative career in science that it is next to impossible to achieve distinction without at least some of them? The few well known examples of outsiders who have done so counter that conclusion, but their rarity suggests that the hurdles which one must cross to do so are very high."

"After retirement, Krebs himself maintained the same disciplined schedule, appearing promptly in his lab at 8:30 AM, five and a half days per week, that he had carried on throughout his scientific career. He was still actively involved in the activities of his lab until a few weeks before his death, at the age of eighty-one, in October 1981."

Motivations for Krebs: 1) Curiosity, 2) "His ambition to justify his choice to become a scientist in the face of the doubts of his father, his mentor, and himself about his ability to "make a success in this field," and 3) justification vis-a-vis those who support me by putting financial resources and facilities at my disposal."

"Enterprises rarely come singly. The creative person often differentiates a number of main lines of activity. This has the advantage that when one enterprise grinds to a halt, productive work does not cease. The person has an agenda, some measure of control over the rhythm and sequence with which different enterprises are activated."

"Because the various enterprises may entail different levels of difficulty or risk, the person may choose at different times to work on a particular one that fits his mood and needs at that point. Finally, the network of enterprise helps the creative person to define his or her own uniqueness." (some of these ideas are from Howard Gruber, I should note)

"In the investigation that led immediately to the discovery of the citric acid cycle, the most significant unforeseen event was the appearance of the paper by Carl Martius and Franz Knoop that gave Krebs the critical clue concerning the likely pathway of decomposition of that metabolite... It is most probable that he came across Martius and Knoop's paper, at the earliest possible moment of its appearance in the Sheffield library, because he was in the habit of using his spare time to scan through the latest issues of biochemical journals."

"To remain productive an experimental system must be sufficiently open to generate unprecedented events by incorporating new techniques and devices, but sufficiently closed to maintain its reproductive coherence. 'It has to be kept at the borderline of its breakdown'." (says Rheinberger)

"Investigators operate within highly ordered frameworks shaped by all the past work in their domains, and events can be recognized as unpredicted only by their deviation from that which the preexisting order does predict. The unpredicted events which the pursuit of investigative pathways so regularly leads researchers create temporary pockets of disorder, and successful responses to these events most often involve adjusting the previous order at its edges sufficiently to fit the seemingly disorderly result into its texture."

"[Thomas] Kuhn's dichotomy represents, I believe, two ends of a broad spectrum, between which the research of leading investigators often has an intermediate character. Not only the means by which a puzzle is solved, but the nature of the solution is often more surprising and more original than his characterization of such work as "mopping up exercises" would imply."

"Daily thought ordinarily cycles repetitively back on itself, with slight variations on previous thoughts, and only widely spaced modifications representing some degree of novelty. Stable changes in ideas, according to Gruber, evolve at a moderate rate. 'The movement of ideas is far slower than the swift but transitory currents of the continuous stream of thought which serves as the 'carrier wave' of creative work.'"

"In the case of experimental scientists, the rate-limiting factor in their mental progress may be set not by the speed with which such thoughts can course through their brains, but by the pace with which they can translate useful thoughts into laboratory operations."

Krebs "was a disciplined improviser, responding quickly to ideas that he came across often by chance, as well as to unexpected observations, flexible enough to change directions opportunely, consistent enough not to become lost in the welter of possibilities that he encountered."

"The great "moments of discovery" in science have appeared to be intimately associated with powerful, instantaneous experience of illumination that occur when the reorganization of ideas carried out by the unconscious suddenly wells up into the conscious awareness of exceptionally creative thinkers."

HOWEVER, "Grober, who has closely reexamined the nature of Eureka experiences--or, as he sometimes calls them, "Aha experiences"--has argued that such events are more frequent than is often imagined, that they are not instantaneous, but have a structure that develops over a measurable interval of time, that they are less likely to represent complete ruptures with the past than is usually imagined, and that they are linked to events that precede and follow them like the crests of an ongoing wave. From his experience in experimental psychology, he infers that it requires at least several seconds to conjure a familiar image or to recognize a fragmented one, and that to become aware of a train of related ideas of any complexity at all would probably take longer still. The time required is sufficient so that an individual beginning to have an insight is likely to be aware of where it may be headed before it fully emerges, and able to consciously to "steer" it in a preferred direction, or avoid a direction that feels unsafe."

"Memory ordinarily simplifies the past by conflating within the single, most powerful of a sequence of similar experiences, events that in fact had occurred repeatedly," making "aha" moments appear more forceful than they really are. "Ever simplifying, memory regularly suppresses those aspects of a complex progression of events that did not prove, in retrospect, essential to an outcome. Points along such progressions that are accompanied by strong emotions, such as the excitement one may feel when recognizing a new possibility, are apt to overshadow what came before and after under calmer circumstances."

"In every realm of our lives we confront complexity beyond our capacity fully to comprehend, and we are forced to find ways to simplify what we encounter."

Friday, March 19, 2010

The "Old Shit Was Better" Effect In Action

Tim Burton's Alice In Wonderland has largely flopped on imdb--7.0 with 25k votes and likely to fall further. Maybe it will even drop to below average (6.7) territory! But also in the past week, his 1990 movie Edward Scissorhands has been making a surge. It's on the top 250 (albeit #250 exactly) for the first time I can remember, and I've been following this thing for about three years. Anyway, if I'm right and if this generalizes, it suggests that it is not only time but actually the contrast of recent bad work that makes an artist's "old shit" look so appealing.

####

It's such a sad state of affairs that I've been reduced to putting expletives into neutral-distancing scare quotes. My high school self would be like, ashamed.

Thursday, March 18, 2010

Are We Not Having Fun Yet?

I think the most legit objection to my* modest proposal to quantify every single subjective aspect of sports is that it might take some of the fun out of watching. When a football fan looks up Brett Favre's career INT stats, the fan "knows" something that the stats can't tell that has some value in predicting Favre's future and analyzing his past. With this rating system, that knowledge would basically already be a part of his expected value stats, and the fan would have no such "insider" info. And the better the rating system is, the less insider info fans can expect to have.

To the extent that the fun of watching sports comes from thinking you might know more about what will happen than everyone else does (see: the profligate amount of sports gambling, fantasy sports, etc.), it's possible that quantifying player's value like Bill James overdosing on adderall really would make sports less fun to watch. This is possibly related but not exactly the same as why expected value poker would be less fun to play. There is a degree to which the variance in the outcome of a given event is what makes it exciting to watch.

* I hesitate to call it "mine" because I'm willing to be that I'm not the first to have thought of this idea and I'm probably not even the first to write about it, although FWIW I couldn't find anything on a quick Google search.

####

Today and tomorrow are the best two sports days of the year. You can quantify your degree of manliness by looking at the highest regional seed you have advancing to the final four:

Only 1 seeds: Welcome Christiano Ronaldo to AMTB! Who knew you had such a refined taste for psychology and statistics?
Highest is a 2 seed: Why do you even bother pretending to shave in the morning?
Highest is a 3 seed: Your favorite video game is probably Super Princess Peach.
Highest is a 4 / 5 seed: You probably prefer smooth to crunchy peanut butter.
Highest is a 6 / 7 seed: No one is going to call you out for this... to your face.
Highest is a 8 / 9 seed: OK in my book.
Highest is a 10 seed +: Welcome to the upper echelon of manliness. Drink it in... it always goes down smooth.

Wednesday, March 17, 2010

Rating Interceptions Probabilistically

Just as it's hard to emphasize the process instead of the results in medicine, it's hard to correctly weight the role of luck in determining the outcome of a given play in sports. But, especially in sports with small sample sizes like football, it is crucial to isolate just the effect that the player himself had on the play.

One way to do this is by having watchers rate the expected value that a player contributes to a given play. For example, instead of just seeing a binary 0 or 1 interceptions on a given play, a continuous rating system should give more information.

So, give 0.5 expected interceptions for a throw that would have been picked off 50% of the time by an average defense / rest of offense (as wideouts play a role, too), 0.05 for a ball that probably wouldn't have been picked off unless the wide receiver hadn't tipped the ball off his hands, or 0.9 for a ball that was basically thrown right to the defender. Of course these ratings will be rough and subjective. But by and large they will be better than the status quo raw stats.

For example, this interception would be rated as ~ 0.9 expected interceptions, this one would be rated as ~ 0.02 interceptions, and this non-interception would be rated as ~ 0.6 expected value interceptions. You then divide the number of expected value interceptions by the number of passing attempts, possibly regress for the style of offense, and you'll have a much more reliable measure of the propensity of the QB to throw an interception than the raw interception stats.

This is what most scouts already intuitively do. But this rating method quantifies it so that the knowledge is more transparent and accessible. I'm willing to bet that this expected value measure, if adopted, would predict the QB's actual interceptions in the following season better than his actual interception total from previous seasons. Is anyone willing to bet against me on that?

This generalizes to other sports statistics, like the expected outs from an outfielder's throw to a runner speeding to home plate, the expected goals of a shot in soccer, or even the expected points from a jump shot in basketball (i.e., air balls are like 0, and points that hit off the rim more than once before going in are less than a swish). It generalizes beyond sports too, to emphasizing the process over the results as an important part of any performance analysis.

Tuesday, March 16, 2010

The Dark Side Of Altruism

Contrary to some haters, altruistic acts that benefit others at costs to oneself are indeed possible and real. But altruistic acts do not necessarily accrue benefits to all others. Instead, they usually are done to benefit one's own in-group. One extreme example of this "dark side" of altruism is suicide bombing, as explained by Robert Pape and summarized briefly here.

What this underscores is that altruism is not necessarily laudable and can be dangerous in the absence of cosmopolitanism. And in general, I think that altruism + cosmopolitanism = preference utilitarianism. So most of the "good" usually discussed regarding altruism can be "explained away" by the preference for max utility.

Monday, March 15, 2010

Ironic Process In Hurt Locker

At one point in the movie, the camp doc / mental health provider tells one of the soldiers that he has "got to change the record in [his] head. You gotta start thinking about other things. Okay, Stop obsessing... Right now."

Did anyone else start laughing at this? Even if you haven't heard of the ironic process theory that attempting to directly suppress thoughts makes them more persistent, surely you can assume that the soldier has already tried to not think about his obsession (dying) on his own. Instead of this course, some sort of cognitive behavioral therapy would likely have been much more helpful.

####

Hurt Locker is currently rated an 8.0 overall on imdb. 17k US voters have given it a 8.3 average, and 37k non-US voters have given it a 7.8 average. It does best with voters under 18, who give it a 8.6 average. So, does it make you like the movie more or less to know that younger people tend to like it the most?

Sunday, March 14, 2010

Preferring Process Is Tough

In Complications, Atul Gawande writes about how we should prefer process over results, becase,
Good decision can have bad results (sometimes people must take terrible chances), and bad decisions can have good results.
In other words, you should evaluate your decisions on the basis of the knowledge that you had when you made the decision. Be process-oriented as opposed to results-oriented.

This is a noble maxim, but it's important to recognize just how hard it is. For example, elsewhere in the book, Gawande describes the case of a patient whose treatment went awry:
In a lab test done the day before, the patient's clotting had seemed slow, which wasn't serious, but an ICU physician had decided to correct it with vitamin K. A frequent side effect of vitamin K is blood clots. I was furious. Giving the vitamin was completely unnecessary--just fixing a number on a lab test. Both the chief resident and I lit into the physician. We all but accused him of killing the patient.
So they blame the resident for administering vitamin K, which was apparently a bad decision. But when autopsy shows that the vitamin K-induced blood clots were in fact not the cause of death, Gawande writes that,
In the days afterward, I apologized to the physician I'd reamed out over the vitamin.
It is good to admit it when you have erred, but I don't see why Gawande would apologize here. Either the resident administering vitamin K was a bad decision at the time it was made or it wasn't--the specific cause of death, the "result," shouldn't change anything.

I don't mean to pick on Gawande. What this example shows is that even people who write about preferring process over results have a tough time applying it. It is really quite an unnatural thing to do.

####

Gawande's three books are all interesting, but Better is the most data-deluged (a good thing) and focused. Checklist Manifesto should have cut the fluff and focused more on the interesting explanation that checklists have yet to be adopted despite yielding huge results because they are low status.

Friday, March 5, 2010

Of Course Aging Research

When I first read Katja Grace's theory that phrases like "clearly" or "of course" are usually used to cloak messy logic in a shroud of obviousness, I was skeptical. But enough examples have convinced me that she is on to something. For example, consider the end of Perl's interesting article on the characteristics of people who live to 100 years old:
The hope, of course, is that these gene discoveries will help in identification of drug targets and creation of drugs to allow persons to become more "centenarian-like" by maximizing the period of their lives spent in good health.

To me, it is by no means obvious that this should be the goal of comparative gene analysis in centenarians. Other goals could be simply attempting to increase lifespan and hoping that healthspan increases concurrently, or ending aging completely*, or eventually transfecting the lifespan-extending alleles of identified genes into zygotes in utero. Since there are plenty of alternatives, in this case "of course" serves to hide the fact that the author is making a value statement.

* Although, contra de Grey and with the caveat that I am no expert, I think we are quite far away from that.

Thursday, March 4, 2010

Game Theory In School Group Work

Students often complain that they don't like group work because they wind up having to do "all of the work themselves."1 They obviously haven't figured out the game.

Once assigned to a group, your group mates will immediately begin to smell for signs of weakness: the weakness of being willing to do more than your fair share of the work. There are some pretty obvious signals of this, like being the first one to organize a meet-up time or saying that you "really want to do well on this assignment."

Ostensibly the other group members will agree with this, and in fact2 they may even convince themselves that they really want to work hard too. At first. But when it comes to crunch time, why would they bother doing the work when you have clearly signaled that you would ultimately be willing to do it for them?

The key to avoid doing excess grunt work is to counter-signal from the get-go. Don't fool yourself into thinking that you can pressure others to do more work, as you hold none of the carrots. The more earnest you are, the more moral hazard you give to your fellow group mates to do less work. Even if you do want to work hard and do well, don't make a big show of it and give others an incentive to slack off.

Teachers could avoid this problem easily by having students grade one another on their effort during the group assignment. But this is far too unconventional and most teachers wisely worry more about their general perception by students than actual fairness3. So the status quo remains.

1: I.e., the phrase "I hate group work" has 115k Google hits.
2: Their far mode mind system is active here, so they're thinking about their morals and what kind of person they generally want to be, instead of the details that would go into actually doing the work.
3: Teachers do this to minimize the probability of low student ratings at the end of the year, as they are generally risk averse and want to keep their jobs.