Book Review: The Old Man, the Sea and the PhD by Martin Lukačišin

The best thing about The Old Man, the Sea and the PhD is that it’s very short. I read it while travelling to the airport and waiting at the security check. It reads very well and it’s packed with ideas. Each idea is presented using a parable which makes them very easy to remember.

The last chapter likens the life of a PhD student to Hemingway’s The Old Man and The Sea, which is where the name of the book comes from. I found this analogy the best and most memorable.

During my time in academia, I wanted to write similar thoughts in a blog post (or two) but I always deleted what I wrote, since I didn’t believe that academia could be fixed and thus my drafts were pure rants. The author is slightly more optimistic and tries to provide solutions, though he still admits that the book is very negative about academia.

I think academia will never get fixed, since it’s stuck in a bad equilibrium, just like schools, medicine, politics or airport security checks. We think academia is about science, schools are about education, medicine is about health, politics about improving lives of ordinary people and airport checks are about security. However, all of them are primarily about something else and the stated purpose is at best a secondary side-effect.

The real issue with academia—in my opinion—is that it’s primary about academics gaining prestige and status among other academics. Science ends up being a secondary side-effect of this process. The author hints at it in the very last paragraph of the book but unfortunately doesn’t expand on the idea. For this reason, I recommend complementing The Old Man, the Sea and the PhD with Elephant in The Brain by Kevin Simler and Robin Hanson. You’ll have a better understanding why the life of a PhD student is as miserable as Martin Lukačišin describes it.

Note: this review has also been published on Amazon.

How I got much better

In the last 18 months, I stopped going to doctors and instead focused on self-experimentation. Many experiments failed, but overall it’s been a big success. Some symptoms are completely gone, some are still alive but few and far between, and some are thriving as well as for the last 15 years.

This post was supposed to be titled Living with histamine intolerance, part 4, but I don’t classify as histamine intolerant anymore. You can read parts 1, 2 and 3 to get better context, though.

Feel free to get inspiration from my experiments but remember that there are risks involved.

Life-style changes that worked

Sunlight

I always wondered why I felt better in the summer and worse in the winter. Each November my ability to exercise rapidly dropped, except when I spent autumn in sunny California and the symptoms were then postponed by two months.

I’ve tried spending every second in the Sun but it was very difficult to do in Sweden. Spending one week under the Arctic Sun didn’t do anything. Two days of kayaking helped a lot, but not as much as 3 hours of full-body sun bathing. Sunlight is the best cure against insomnia and bad sleep.

On the other hand, taking vitamin D in winter helps but doesn’t bring the full effect. It’s not so surprising, since sunlight does more to our skin than just vitamin D production and we’re only discovering those effects now.

Dietary changes

Diamine oxidase (DAO) is an enzyme that regulates histamine in our bodies. The only macronutrient increasing DAO concentrations in the intestinal lymph in rats is fat. In another study of rats, MCT oil increased DAO activity 2.5 times in the intestines.

Even though I’m not a rat, I had to try it out. Did it help? Maybe it did, maybe it didn’t. It’s true that my DAO levels are 3 times higher than before, but I have too little data to judge causality. An unexpected side-effect was a big decrease in headaches and migraines. This is perhaps not that surprising, since MCTs boost ketone production and ketogenic diets are therapeutic for a few neurological conditions.

The second big change in my diet was eating nose to tail – eating other parts of animals than just muscles. Organ meats are cheap, easy to cook and super-nutritious. Skin, bones and cartilage are rich in collagen, which seems to be beneficial for many. The nutritional wars concentrate too much on carbohydrates versus fat, while different proteins and amino acids are often ignored. That’s likely a big mistake.

I’ve changed other small things in my diet, but listing them all would need another full post. Luckily for you, someone made a nice infographic and wrote a book about my diet 4 years ago. After I found out about people copying my diet, I read the book and had a lot of Aha! moments. That being said, the book is overconfident at times and I’d be extra careful following their advice to the last word. Also check Scott Alexander’s critical review.

Probiotics

Abdominal pain has been my number one problem for years. I’ve consulted doctors and tried many things, but those brought minor improvements at best. With Lactobacillus plantarum, the pain was gone within hours. Bacillus coagulans was also helpful for better digestion, but I had to take it every day and was later able to achieve a superior effect with glycine-rich foods, so I stopped taking it.

While I swear by L. plantarum, I do believe that probiotics are mostly unnecessary. If you have to take them every day, there is probably a deeper problem that needs addressing.

Intermittent fasting

I’ve previously written about two stressors, cold showers and exercise, helping me cope with various symptoms. At about the same time I’ve started experimenting with another stressor: intermittent fasting. Surprisingly, fasted state is when I have the best focus and productivity. I usually skip breakfast and only eat within an 8–9 hour window, with a big lunch and a big dinner (or two).

Fasting is a stressor that requires caution, but an intuitive approach works well for me. When the body stops telling me “This is uncomfortobale, but I like it” and switches to “Are you trying to kill us?”, it’s time to listen and eat something.

Supplements

I don’t like using supplements, because I often forget to take them. However, they are a great tool for controlled experiments. For example, I’ve taken pure glycine for two weeks with great results and after that incorporated a lot of collagen into my diet (with the same results).

It’s all just placebo and correlation doesn’t imply causation!

If everything I’ve experienced is a placebo effect, where was it for the previous 4 years? Out of the 50+ lifestyle changes, prescription pills, supplements and probiotics I’ve tried, only about 15 worked. If there was placebo effect, it must have been negligible. However, the good ol’ regression to the mean (patients getting better on their own) definitely played a role. Then again, I’ve tried relying on it many times before and it was never sufficient on its own.

How do I know the interventions listed above helped? Either I’ve done sufficiently many experiments or the effect was so strong it’d be impossible to explain otherwise. For example, after taking nonsteroidal anti-inflammatory drugs (NSAIDs) regularly, strong abdominal pain started. Wikipedia lists gastrointestinal problems as the main adverse effect of NSAIDs and they are also known to disrupt gut flora. Nothing has helped me for years. Then one day I tried Lactobacillus plantarum and the pain was gone within hours. Fixing other stomach problems took another 10 months, but at least the pain was gone. Also, L. plantarum is one of the few probiotics that performs in medical trials on people and in other studies, animals benefited from probiotics after being given NSAIDs.

A true skeptic would say that correlation doesn’t imply causation and that such evidence is not sufficient. Many people don’t believe successful stories and dismiss any kind of evidence, often because it contradicts their beliefs.

Skeptics are right that no evidence is perfect and we can never be completely sure. I admit I didn’t use an identical twin as a control group and I didn’t double-blind my fasting experiments. However, my personal experiments showed that skeptics are as likely to be wrong as the people they criticize. I’m glad I didn’t listen too much to sites like QuackWatch and Science-Based Medicine, otherwise I would probably avoid probiotics, cholesterol and fat for no good reason.

The new skeptic movement reminds me of the disagreement between academic skeptics and Pyrrhonian skeptics 2 millennia ago in Greece. Academic skeptics claimed “Nothing can be known, not even this” while Pyrrhonians viewed them as too dogmatic. I’m siding with Pyrrhonians here and my advice is: be skeptical, especially of skeptics.

This post is getting too long, so let’s take a short break with classical music.

Self-experimentation and its risks

A while back my mom told me to drink a particular herbal tea to alleviate some of my symptoms. I replied with “If this herb was effective, it’d have performed well in scientific trials and doctors would have prescribed it to me.” The statement does make sense, but it involves many strong assumptions and often isn’t true.

As an example, there is a drug called ramelteon, which binds to the melatonin receptors and for all we know it has identical effects to melatonin and performs about the same as melatonin in clinical trials. A monthly supply of melatonin is about $5 while ramelteon costs $100, yet doctors prescribe ramelteon instead of melatonin. Or one time I asked a Swedish gastroenterologist about her opinion on L. plantarum and other probiotics. She told me I know more than her and I should keep trying. This is despite plenty of successful clinical trials, many done on Swedish universities.

The previous two examples were cures that proved themselves in trials. Using fat and MCT oil in particular to increase the production of DAO has only been tested on rats. Perhaps it also works on people, but as far as I know no experiment was done on people, so doctors wouldn’t recommend it.

Medicine seems to miss a lot of treatments, so self-experimentation might be worth it. It does bear a lot of risks, so I evaluated possible adverse effects and monitored my health. For example, I’ve closely monitored my cardiovascular risk factors and they stayed the same or improved. However, I don’t trust these tests completely. A comprehensive heart-disease blood test would include 5–7 different blood values, but despite so many numbers, its predicting power is limited. I believe that there is too much focus on lipoprotein particle size and soon a more important risk factor will be discovered, so I’m not pronouncing myself completely out of risk yet.

Blood tests are tricky to interpret for another reason. Some blood reference intervals are calculated so that exactly 95% people fall inside them and it’s not always clear if sick people are included or not (see here or here). Unfortunately, many doctors don’t know that and would diagnose you sick or healthy solely based on these arbitrary numbers. It reminds me of a TV discussion with someone complaining that poverty in the US has stayed exactly at 20% of the population for the last 100 years. This came as a big surprise, so the moderator asked about their definition of poverty. “It’s the poorest 20% of the population”, the first person said.

So I take blood tests and reference intervals with a grain of salt. Perhaps more important is tracking values over time and comparing them with a younger version of myself.

Disclaimer: While I don't have a lot of trust in medicine and doctors anymore, a world with doctors is much better than a world without them. Doctors are awesome at fixing anything acute or live-threatening. Next time I break my leg into six parts while skiing, I'm sure doctors will do a great job. It's the chronic conditions where my trust disappears.

Further reading

I recommend 3 blogs for anyone interested in health and medicine. And one crazy idea.

• Whole Health Source: Stephan Guyenet is a neurobiologist blogging mostly about obesity but also other topics. He was one of the scientific advisors of the Perfect Health Diet book, but he would disagree with parts of it and his diet is slightly different, too.
• slatestarcodex.com/tag/medicine: Scott Alexander—the best blogger on the Internet—is a psychiatrist by training and often blogs about problems in medicine, including drug discovery, life-cycle of medical ideas and how skeptics exaggerate their claims. This blog hasn’t affected my life-style but it pointed out the common problems and gave me arsenal to evaluate medicine.
• Mosaic: A non-profit website with awesome scientific journalism about anything life related. Again, I learned a lot from this website but it hasn’t affected my life-style.
• If you like crazy ideas, Robin Hanson suggests to cut health care spending in half, since spending on health care doesn’t correlate with health outcomes anyway. I’m not buying it yet but the idea is worth pondering, as are all Robin’s ideas.

I’ve also read other websites and watched a lot of YouTube videos, but it’s a minefield that is very difficult to navigate. Especially nutrition discussions are often worse than politics or gender, with people citing facts out of context, ignoring evidence and sticking to dogmas. For example, many people benefit from vegetarian or low-carb diets and then preach them as the healthiest diets for everyone. Perhaps they are helpful in certain situations, but that doesn’t mean everyone should follow them or that they are safe in the long term. And some people don’t benefit from any diet, so they become radical skeptics and question everything. Usually there is some truth in each dogma, but some people take them too far.

Maybe this post is also full of untrue dogmas, misrepresented scientific findings and without contradicting evidence. Only time will tell. I tried to do my best, though.

Efficient Use of Exponential Size Linear Programs

After being sick on and off for a couple of years, I decided not to finish my PhD. I did write a small thesis, though. Thus on March 20, 2015, there were two big cosmic events: a partial solar eclipse and I defended a licentiate degree. It was cloudy in Stockholm, so the eclipse was a disappointment, but my defense wasn't and I successfully defended.

I've tried to make the thesis introduction easy to understand and you can now read it here on my blog.

What is the thesis about?

Computer Science is no more about computers than astronomy is about telescopes. – (probably) Edsger Wybe Dijkstra

When Santa Claus needs to find a route around the Earth to distribute all the gifts to children in the shortest possible time, it might take millions of years to compute even if he used all the supercomputers that are currently on Earth. This is an example of a problem in computer science that is not known to be solvable fast even after decades of research.

Maybe we have just not been intelligent enough to find the right approach for this problem. However, if you ask a complexity theorist, they would most likely tell you that it will never be possible. Right now we are not completely sure about this fact, but the evidence is piling up (see the excellent essay The Scientific Case for P≠NP by Scott Aaronson).

If it really is impossible for Santa Claus to find the shortest route to distribute all gifts, what other options does he have? Instead of insisting on finding the shortest route, he might try to find a route that is at most 50% longer than the shortest one. Arguably, this should be an easier task and traveling a 50% longer route could still allow him to distribute all presents in time.

First Attempt

The problem that Santa Claus needs to solve is called the Traveling Salesman Problem (TSP). The famous algorithm by Christophides [Chr76] finds a solution to TSP within an error of 50% from the optimal solution. This is an example of an approximation algorithm, since it does not solve a problem optimally but only approximately, that is within a certain error from the optimal solution.

The time complexity of this algorithm is n^2.373 operations [MS04] where n is the number of stops that Santa makes. So a computer running the algorithm makes at most about n^2.373 operations.

To get the value of n we need to estimate the number of families with children. According to Gap Minder [Gap11], there are about 1.9 billion children on Earth. The average number of children per family has been estimated to be almost three [Fam06], so we can estimate that Santa Claus will need to make at most 650 million stops. Not all children expect to be visited by Santa Claus but a large fraction does, so this estimate is good enough for our purposes.

If Santa Claus had access to the fastest supercomputer Tianhe-2 (as of November 2014) with the speed of 33.86-petaflops, which is about 3 ⋅ 10¹⁶ operations per second, it should take a couple of minutes or hours to finish the computation. We note that the theoretically optimal subroutine taking about n^2.373 [MS04] operations might in this case be slower than a subroutine that uses Edmond’s algorithm [Edm65] that needs at most n^2.5 operations. In any case, such an algorithm will finish its execution in a couple of days.

A possible problem is the high memory consumption of these algorithms, but for the sake of simplicity of this text let us ignore this problem. A bigger problem might be the 50% error, because a tour that is 50% longer than the shortest one might be too long to distribute all the presents in time.

Improving the Error

The celebrated algorithm by Arora and Mitchell [Aro98, Mit99] can find a solution to TSP within an error of say 20, 10 or 5% from the optimal solution. We can pick any error margin we like, but the smaller the error the longer the running time will be. The algorithm works provided that the problem is situated in a Euclidean space. For the purpose of this text, we only need to know that the well-known 2-dimensional (2D) plane or 3-dimensional (3D) space are both Euclidean.

We could pretend that the Earth is flat by using one of the many map projections. Unfortunately, map projections do not preserve an important geometric property, namely that the space is Euclidean and thus cannot be used by the algorithm.

Let us then try using 3 dimensions. The Arora-Mitchell algorithm assumes that the shortest path between two points is the straight line connecting them, but this is a problem for Santa, since he cannot travel under the ground. As it turns out, it is not such a big problem in the end.

Santa Claus can run the Arora-Mitchell algorithm on all positions of children on Earth as if they were simple points in a 3-dimensional space. Then he follows the route returned by the algorithm with the important change that he flies over the Earth’s surface instead of following the straight line under ground. How much longer is the route when flying over the Earth’s surface?

The most remote inhabited place on Earth, Tristan da Cunha, is about 2000 kilometers from the nearest inhabited land. However, the direct distance through the Earth is about 11.5 kilometers shorter, which is only about 0.5% smaller. The situation is shown on the picture below.

The distance between Tristan da Cunha and the nearest inhabited place Saint Helen is about 2000 kilometers. The direct distance through the Earth is about 11.5 kilometers shorter, which is only about 0.5% smaller.

Of course we cannot be sure that 2000 kilometers will be the longest distance between two stops found by the algorithm, however it will most likely not happen very often in the optimal tour. Human settlements are fairly spread out over the Earth’s surface and intuitively an optimal tour should not make a lot of long jumps—it is likely more efficient to first distribute all gifts in Asia and then go to Europe rather than flying 2000 kilometers back and forth all the time.

We thus have a strong reason to be optimistic, so let us hope that the Arora-Mitchell algorithm returns a tour such that Santa travels only about 0.5% longer route compared the one following straight lines under ground. The Arora-Mitchell algorithm is able to return a solution within x% of the optimal tour for any error x we choose. How big should x be if we want to get a 10% error?

If your answer is 9.5%, you are close but not completely right. We want to have 1.005 ⋅ (1 + x) = 1.1, so x ≈ 9.45%. Let us now estimate how long the algorithm will run. The number of operations performed by the algorithm in 3 dimensions is at least

where n is the number of families to visit and a is a number that is at least 9 [RS98]. We note that the exact value of a is higher than 9, but for our purposes this information is enough.

Even if Santa Claus had access to the fastest supercomputer Tianhe-2, it would take about 10¹²⁵⁰ years to run the algorithm (10¹²⁵⁰ is a number with 1250 zeros at the end). Or equivalently about 10¹²⁴⁰ times the age of the universe. Even if each person on Earth had such supercomputer and we ran the algorithm in parallel on all the supercomputers, the running time would still be 10¹²³⁰ times the age of the universe.

For comparison, the Held-Karp algorithm [HK62] that makes about n²2ⁿ operations is the fastest known exact algorithm that finds the optimal route. It would run 10^100000000 times the age of the universe. The running time of the Arora-Mitchell algorithm is tiny compared to this one.

Heuristics

Even though the problem that Santa Claus faces may sound artificial, it appears in real life for companies that distribute goods. Instead of approximation algorithms, they often use heuristics, which are computer programs that try to solve a problem but only weak guarantees are known about the quality of their output.

For an approximation algorithm, we always have a guarantee that the output will be of a certain quality, but we might have to pay for this guarantee with slower running time, as in the case of the Arora-Mitchell algorithm. Heuristics, on the other hand, are usually very fast but we might be unlucky and receive a solution that is not good enough.

In 1947, Dantzig [Dan51] developed the simplex method that became one of the most important algorithms of the 20th century. It performs very fast in practice but for many years its theoretical speed was unknown. In 1970, Klee and Minty [KM72] found some examples where the algorithm became very slow. The algorithm continued to be used in practice and still is until today. It used to have a similar status as heuristics, since we did not know when and why it performs well.

In 1977, Borgwardt [Bor86] found a class of linear programs that are provably solved quickly by the simplex method. Subsequently, other people have found cases where the simplex method performs well. Finally, in 2001—five decades after the algorithm was discovered—Spielman and Teng developed a powerful tool called smoothed analysis [ST04] that explained why the simplex method works fast in practice while there are rare cases where it is very slow.

For many real-world problems, heuristics are both faster and better performing than approximation algorithms and we do not know why, as used to be the case for the simplex method. It is not unheard of that heuristics achieve 1% error as well as they are very fast, while the best-known approximation algorithm only achieves an error in the order of tens of percents (see for example [JM07]). Some people use this fact to argue against the use of complexity theory, but we would argue the exact opposite. Since some heuristics seem to perform well, we should study them more and try to figure out why they perform so well or find cases where they do not, as we did in the case of the simplex method. If anything, this is an argument for more research in complexity theory, not less.

What Should Complexity Theorists do?

There are several interesting research directions that help us. For example, we can try to find faster approximation algorithms for which we do not have to wait 10¹²⁵⁰ years to finish. We can also try to prove that some heuristics used in practice perform very well, hence turning a heuristic into an approximation algorithm. The hardest alternative of all is to prove that there will never be an algorithm that finds good solutions and is fast at the same time. Such negative results are called hardness results, because they prove that a problem is hard to solve fast.

History has shown that it is important to be both optimistic and skeptical at the same time. Some problems were thought to be hard but an algorithm was found in the end; while for other problems, a very simple and naive algorithm turned out to be the best we can hope for. However, for most of the problems, their complexity is still not well established. The best-known algorithms are often very far from the best-known hardness results and the truth can lie anywhere in between.

What is the Contribution of this Thesis?

In the past decades, linear programming (LP) has been successfully used to develop approximation algorithms for various optimization problems. The simplex method mentioned earlier is used to solve linear programs and is considered to be one of the most important algorithms of the twentieth century, which underlines the importance of linear programming in computer science.

The so-called assignment LP lead to substantial progress for various allocation problems, where items are allocated to players or machines. An example of such problem is scheduling unrelated parallel machines, where computer tasks are to be scheduled on a number of machines such that the last one is finished as soon as possible. Unfortunately, we know that the best-known approximation algorithms for this problem already have the best approximation ratio that any algorithm using the assignment LP can achieve. The bound on the achievable approximation ratio by an LP is called the integrality gap of the LP.

We have thus reached the limits of the assignment LP. The natural question is then whether a different LP formulation can lead to better algorithms. We answer this question positively for variants of two allocation problems: max-min fair allocation and maximum budgeted allocation, which we define in later chapters. This is achieved by using a more powerful LP formulation called the configuration LP that has exponential size. Exponential size is considered inefficient, but configuration LP can be approximated efficiently. This is where the title of the thesis comes from, since we use exponential linear programs in an efficient way.

The restricted max-min fair allocation, also known as the restricted Santa Claus problem, is one of few problems that has a better polynomial estimation algorithm than approximation algorithm. An estimation algorithm estimates the value of the optimal solution, but is not necessarily able to find the optimal solution. The configuration LP can be used to estimate the optimal value within a factor of 1/(4 + ε) for any ε > 0, but it is not known how to efficiently find a solution achieving this value. We give an approximation algorithm with the same approximation guarantee but improve the running time to n^{O(log n)}. Our techniques also have the interesting property that although we use the rather complex configuration LP in the analysis, we never actually solve it and therefore the resulting algorithm is purely combinatorial.

For the maximum budgeted allocation (MBA) we prove that the integrality gap of the configuration LP is strictly better than 3/4 and provide corresponding polynomial time rounding algorithms for two variants of the problem: the restricted MBA and the graph MBA. Finally, we improve the best-known upper bound on the integrality gap for the general case from 5/6 ≈ 0.833 to 2√2 − 2 ≈ 0.828 and also show hardness of approximation results for both variants studied.

Further reading

The full thesis is available for download and you can also have a look at my slides, which contain plenty of additional pictures.

Bibliography

[Aro98]    Sanjeev Arora. Polynomial time approximation schemes for euclidean traveling salesman and other geometric problems. J. ACM, 45(5):753–782, 1998.

[Bor86]    Karl Heinz Borgwardt. A Probabilistic Analysis of the Simplex Method. Springer-Verlag New York, Inc., New York, NY, USA, 1986.

[Chr76]    Nicos Christofides. Worst-case analysis of a new heuristic for the traveling salesman problem. Technical report, CMU, 1976.

[Dan51]    George B Dantzig. Maximization of a linear function of variables subject to linear inequalities, in activity analysis of production and allocation. 1951.

[Edm65]    Jack Edmonds. Paths, trees, and flowers. Canadian Journal of mathematics, 17(3):449–467, 1965.

[Fam06]    Number of people in a family. http://web.archive.org/web/20140819085940/http://hypertextbook.com/facts/2006/StaceyJohnson.shtml, 2006. Accessed: 2014-08-19.

[Gap11]    The world has reached peak number of children! http://web.archive.org/web/20141120230405/http://www.gapminder.org/news/world-_peak-_number-_of-_children-_is-_now/, 2011. Accessed: 2014-11-20.

[HK62]    Michael Held and Richard M Karp. A dynamic programming approach to sequencing problems. Journal of the Society for Industrial & Applied Mathematics, 10(1):196–210, 1962.

[JM07]    David S Johnson and Lyle A McGeoch. Experimental analysis of heuristics for the stsp. In The traveling salesman problem and its variations, pages 369–443. Springer, 2007.

[KM72]    Victor Klee and George J Minty. How good is the simplex algorithm? 1972.

[KMN⁺14]   Christos Kalaitzis, Aleksander Madry, Alantha Newman, Lukas Polacek, and Ola Svensson. On the configuration lp for maximum budgeted allocation. In IPCO, pages 333–344, 2014.

[Mit99]    Joseph S. B. Mitchell. Guillotine subdivisions approximate polygonal subdivisions: A simple polynomial-time approximation scheme for geometric tsp, k-mst, and related problems. SIAM J. Comput., 28(4):1298–1309, 1999.

[MS04]    M. Mucha and P. Sankowski. Maximum matchings via gaussian elimination. In Foundations of Computer Science, 2004. Proceedings. 45th Annual IEEE Symposium on, pages 248–255, Oct 2004.

[PS12]    Lukas Polacek and Ola Svensson. Quasi-polynomial local search for restricted max-min fair allocation. In ICALP, pages 726–737, 2012.

[RS98]    Satish B Rao and Warren D Smith. Approximating geometrical graphs via “spanners” and “banyans”. In Proceedings of the thirtieth annual ACM symposium on Theory of computing, pages 540–550. ACM, 1998.

[ST04]    Daniel A. Spielman and Shang-Hua Teng. Smoothed analysis of algorithms: Why the simplex algorithm usually takes polynomial time. J. ACM, 51(3):385–463, 2004.

Evolutionary psychology: understand culture and improve your life

Why do cultures in Europe differ a lot from north to south, but much less from east to west? Does it have something to do with temperature? Why are societies that are surrounded by more bacteria and viruses less tolerant and more xenophobic? And why are symmetric objects considered beautiful all over the world?

To answer these questions, we need to look into our evolutionary makeup. The mind is not just a product of culture, but also a product of evolution and environmental factors other than culture.

Massive modularity of our brains

Many families have children as well as pets, for example a dog or a cat. Even though the pets and children live in the same environment, only one species learns the language spoken in the family. Some people claim that it's because of an all-purpose learning device in our brains, but that is false (see here and here). We have pre-programmed mechanisms (also called modules) in our brains that are designed to learn a language.

We can think of the brain as a computer that is loaded with many useful programs and libraries that learn from the environment. We have mechanisms for face recognition, short-term memory, naive physics, cheater detection, incest avoidance, naive psychology, food disgust, general intelligence, etc. If you lack a module in your brain, you can't circumvent it by learning; the same way you can't fly or breathe under water, because you lack important body parts. Cats and dogs won't be able to talk to us unless their brain evolves and after a stroke you might lose the ability to talk coherently. Malfunctioning brain mechanisms are also the root causes of disorders like Down syndrome, Asperger syndrome and prosopagnosia.

Rider on a elephant

A very useful metaphor about our brains comes from Jonathan Haidt's excellent book Happiness Hypothesis. Our brains consist of a rider controlling an elephant (Daniel Kahneman uses System 1 and System 2 for essentially the same thing). The elephant is responsible for our emotions, intuition, bodily functions, etc. The rider is responsible for rational thought, self-control, etc. The elephant has very old parts that we inherited from reptiles and mammals, while the rider is very young and is also present in higher mammals.

Ironically, the rational and younger rider thinks that he has full control, while it's often the elephant who makes the big decisions. The rider has only a limited power against the strong elephant. Imagine you heard about an excellent work opportunity, but later decided that commuting 10 more minutes to work is not worth more salary and better work conditions. The elephant was very scared and wasn't going anywhere, so the rider came up with his own story where he has everything under control.

The first principle is that you must not fool yourself—and you are the easiest person to fool. – Richard P. Feynman

Nature versus together with nurture (and randomness)

You know the classic question: is behavior X determined by biology or is it determined by the environment? This question gives the impression of there being only two alternatives, a notion now recognized as a false dichotomy. Robert Sapolsky said it well.

To an overwhelming extent, the age-old “nature versus nurture” debate is silly. The action of genes is completely intertwined with the environment in which they function; in a sense, it is pointless to even discuss what gene X does, and we should consider instead only what gene X does in environment Y. – Robert Sapolsky in Peace Among Primates

An important, but often overlooked, factor is randomness. We have chemical processes running in our bodies that are inherently random and those random events affect our behavior. There is also randomness in people's lives—two people will never go through the same experiences. Randomness is one of the reasons why identical twins raised by the same parents in the same environment end up being slightly different. Another overlooked factor is prenatal environment. What your mom ate and whether she was stressed out during pregnancy affects your behavior forever.

An example of how all this plays out is our status-seeking behavior. We all have a brain mechanism that scans the environment around us, measures our own social status and tries to increase it. Entrepreneurs improve status by starting a successful company, professors by getting their papers published and a hunter-gatherer perhaps by killing a big animal. A basic behavior is provided by nature and then culture and environment shape it into its final form. Note that there are also gender differences, men and women on average seek status differently.

Many people have small jaws these days and therefore problems with teeth not fitting in. It is unlikely that this is caused by bad genes, since only 10 000 years ago people had bigger jaws with better teeth. Scientists theorize that it was because they chewed harder food, while these days it's mostly soft food, so the jaw doesn't develop properly. Our brain is much like the jaw. Overprotected kids suffer a deficit of play and they grow into less creative, less social and less happy adults. Even if a behavior is influenced by genes, the genes need input from the environment for the behavior to develop properly. Nature and nurture are truly inseparable.

Evoked and transmitted culture

Have you ever wondered why nations in colder areas have tasteless food? The classic explanation attributes this to cultural differences. This politically correct answer is completely useless: cultures are different because they are different (although it would be appreciated in the tautology club).

People in warmer regions used spices, because they had to. The commonly used spices have very good antimicrobial properties, so they protect food from spoiling when it's too warm.

Let's now move on to a much more difficult topic than food, namely xenophobia. What you read next might be upsetting, so watch a video of a cute duck first.


Our brains are also pre-programmed to be xenophobic and intolerant, but it seems to be very dependent on the environment around us. Suppose you are a hunter-gatherer living in the mountains. The spring and summer were surprisingly cold and the food is scarce, so you descend from the mountains where you meet another tribe. The tribe has been living in an environment with different pathogens (bacteria, viruses, parasites), so you are not immune to them but they are. Exposure to those pathogens might harm you and even kill you. What might evolution do? It makes you avoid strangers, because they might carry dangerous pathogens. It also makes you avoid new things and be more conservative, because your old ways of living are proven to be safe. Remember that evolution is an amoral process that only cares about transmission of genes, so it sometimes leads to such unpleasant results.

There is also good news, though. People living in areas with fewer pathogens are less xenophobic, more tolerant and open to new experiences. If you live in a safe environment, it's less likely that a contact with a stranger or a new experience will harm you. Rather than avoiding strangers, it's more beneficial to meet them and perhaps trade goods with them.

All of this sounds reasonable, but did we really evolved to be this way? One can do experiments to confirm the theory. In one experiment, people that heard news about a recent disease outbreak were less open to immigration. Unfortunately, it's hard to do experiments with whole cultures, but we can wait for a natural experiment: different parts of the world get rid of pathogens quicker than others, so we can compare how their levels of xenophobia change over time.

Wait, there is more! Countries with lower prevalence of pathogens also value physical beauty less, have puberty earlier, are more individualistic, have higher IQ, are less violent, are more democratic, have fewer kids, etc. The list is never-ending and it remains true even after controlling for factors like wealth and temperature. Prevalence of pathogens might be the most important factor behind cultural differences. It will take decades to prove causality and not just correlation, but it's very intriguing nevertheless.

In the last 100 years, humanity made a lot of progress; many countries ended racial segregation, gave voting rights to women and are now even considering gay marriage. What if this increased tolerance happened due to progress in medicine, vaccination and better hygiene? This social change is often attributed to our rational selves, but perhaps we only tamed the elephants in our brains.

Other environmental factors that correlate strongly with cultural differences include: temperature, abundance of resources, ratio of men and women, whether they have a nomadic lifestyle, etc. When elements of culture are shaped by the environment, they are called evoked.

People in warmer cultures don't need to use spices anymore, because they have fridges, but such recipes are still used by new generations. Old information is being transmitted, so we call it transmitted culture. It is unclear what elements of culture are evoked and which are transmitted, though. The understanding of our brains is limited, so anyone claiming they completely understand culture is lying.

Improve your personal life

We could apply evolutionary theories to social problems, but I like to use them to improve my own life. Most people who tried public speaking felt nervous about it. Someone then told them to just relax and not think about it. However, this well-intentioned advice only focuses on the rider and completely ignores the elephant in the brain.

What if I told you that a better strategy is to stand straight instead? Your body will lower the level of stress hormones, the elephant will feel less scared and the rider will calm it down easily.

Other life improvements are about the new media: celebrity tabloids, video games, TV, novels, movies and porn. Recall that our genes react to the environment around us. What happens if you spend your free time fighting fictional creatures or watching lives of attractive people that are 5 000 kilometers away? Your behavior in the actual culture you live in will be inappropriate and suboptimal.

From Stuff No One Told Me

I'm not saying you should avoid all the new media completely. For example I read novels and watch movies from time to time, but I try to pick the more realistic ones. However, you should avoid porn and celebrity gossip like the plague. On the other hand, feel free to gossip about the people around you or have sex with them—science has shown that both behaviors are beneficial.

There are plenty of other ways to apply evolutionary theories to your life and if you want to know more, see the reading list at the end of this post. The possibilities are truly endless: you can improve your happiness, make better decisions, lower your stress, become a better writer, become healthier, be more efficient or understand your own gender better.

New theories

Evolutionary psychologists have come up with interesting theories that try to explain many facts of human existence.

• Why people gossip.
• Why socialism works in a family or a small community, but ends in a disaster when applied to a whole nation. As E. O. Wilson said, "Marxism: wonderful theory, wrong species".
• Why toddlers are almost indistinguishable from each other.
• Why women put on make-up and why men take large risks.
• Why parenting style has almost no effect on a child's personality (parents affect child's personality through their genes instead).
• Why a small fraction of people is born left-handed.

What if some of the theories turn out wrong? We move on and try to find better explanations. However, some of the theories have already amassed substantial evidence, so they are very likely to be correct.

Interestingly, scientists have failed to provide good evolutionary explanations for humor, male homosexuality, female orgasm and music. Music might be just a by-product of evolution that presses the right buttons all over the brain. Think of it as a pleasurable drug.

Some (flawed) arguments against evolutionary psychology

If behaviors like aggression, psychopathy and xenophobia are encoded in our genes, people think this will give others excuses. You can't kill a person, blame it on your genes and skip going to jail. Since the science can be misused, some people try to stop this line of research. Their flawed reasoning reminds me of my favorite quote.

Numerous studies have shown that at least 98% of murderers were born after a heterosexual intercourse. That's why the European Union should ban people from having sex. – Martin Malý

We could ban sex to end all human suffering, but it would come with at a big price: all the good things human would be gone too.

Mentioning evolutionary psychology or neuroscience often leads to very heated discussions, comparable to those about politics and religion. The new scientific findings affect every single mainstream ideology; it doesn't matter whether you are on the political left or right, an anarchist, a feminist, a Christian or even a vegetarian. Each ideology got something wrong about our minds and bodies.

Some people devoted their lives to an ideology, built their whole identity around it and it's hard for them to admit they were wrong the whole time. Cognitive dissonance then kicks in and people get upset. Many arguments against evolutionary psychology have more to do with emotions rather than objective facts.

Rationality is costly. It prevents us from believing what we want to believe. – Michael Huemer in The Irrationality of Politics

Conclusion

It has been 150 years since Charles Darwin's book On the Origin of Species and people are slowly accepting that evolution is real. However, most people have still trouble accepting that evolution affects our brains too. Many religions have accepted evolution and it's time for the rest of the world to do the same.

I would like to end the post on a positive note, so let me quote Daniel Dennett.

Every living thing is, from the cosmic perspective, incredibly lucky simply to be alive. Most, 90 percent and more, of all the organisms that have ever lived have died without viable offspring, but not a single one of your ancestors, going back to the dawn of life on Earth, suffered that normal misfortune. You spring from an unbroken line of winners going back millions of generations, and those winners were, in every generation, the luckiest of the lucky, one out of a thousand or even a million. So however unlucky you may be on some occasion today, your presence on the planet testifies to the role luck has played in your past. – Daniel Dennett in Freedom Evolves

Further reading and watching

Videos and lectures

The easiest way to start with evolutionary psychology is by watching videos and lectures. It's possible to watch them 25% faster by using the gear button on YouTube.

• Conversation between David Buss and Richard Dawkins. Evolutionary psychology should just be named psychology.
• Conversation between Steven Pinker and Richard Dawkins. Great examples of adaptations.
• Human Behavioral Biology course on Stanford by Robert Sapolsky.
• The Uniqueness of Humans, by Robert Sapolsky. We think that we are the most unique species on Earth but we are actually not that unique.
• Darwin's Strange Inversion of Reasoning, by Daniel Dennett. A beautiful connection between biology, evolution and computer science: we are made of trillions of mindless little robots.
• Stress, Portrait of a Killer. A document about stress, its origins and consequences. Stress works the same in animals as in us, so we can learn a lot from animals.
• A Darwinian theory of beauty, by Denis Dutton. Why do we consider some things more beautiful than others and why are many things considered beautiful all around the world?
• Richard Dawkins' lecture on natural selection and evolutionary psychology.
• Moral behavior in animals, by Frans de Waal, one of the biggest experts on primates.
• Hjernevask, a 7-part Norwegian documentary covers a lot of the new research and features many prominent scientists.

Books

These books take into account evolutionary psychology and neuroscience. Many of them aren't solely about evolution, but they are based on the newest research.

• The Blank Slate: The Modern Denial of Human Nature, by Steven Pinker. If I were to write one of my old blog posts again, I would name it "7 non-fiction books that changed my life" and Blank Slate would be there. FAQ about the book.
• The Happiness Hypothesis: Finding Modern Truth in Ancient Wisdom, by Jonathan Haidt. The best book ever written. It's about happiness but contains a lot of great wisdom.
• Survival of the Sickest: The Surprising Connections Between Disease and Longevity, by Sharon Moalem. Sometimes a sickness might help you survive. Fascinating journey into medicine, biology, evolution and genetics.
• Is there anything good about men?: How Cultures Flourish by Exploiting Men, by Roy Baumeister. Yes, there is something good about men.
• The Naked Ape: A Zoologist's Study of the Human Animal, by Desmond Morris. Even though we think of human species as superior to everything else, underneath everything are we still animals. This is a very old book and it's interesting that many theories have been falsified since then.
• Predictably Irrational: The Hidden Forces That Shape Our Decisions, by Dan Ariely. Our brain has a lot of heuristics that helped us throughout the evolution but sometimes lead to bad results.
• Why Beautiful People Have More Daughters: From Dating, Shopping, and Praying to Going to War and Becoming a Billionaire by Alan Miller and Satoshi Kanazawa. This is the only book in this list that I haven't read, but it should cover all the basics of evolutionary psychology and its applications.
• The Paleo Manifesto: Ancient Wisdom for Lifelong Health, by John Durant. Improve your health by doing things more in line with our evolutionary makeup.
• Brain Rules: 12 Principles for Surviving and Thriving at Work, Home, and School, by John Medina. Presenting recent advances in neuroscience in an easy way.
• Quiet: The Power of Introverts in a World That Can't Stop Talking, by Susan Cain. Sometimes evolution leads to two different strategies. Introverts are not worse than extroverts, they are just different.
• Drive: The Surprising Truth About What Motivates Us, by Daniel H. Pink. I never thought that motivation would be so tricky.
• Thinking, Fast and Slow, by Daniel Kahneman. About two systems in our minds working at different speeds.
• The Power of Habit: Why We Do What We Do in Life and Business, by Charles Duhigg. How habits are formed and how to get rid of the bad ones.

Some interesting papers

If you feel adventurous, you can read some scientific papers.

• Evolutionary Psychology: Controversies, Questions, Prospects, and Limitations by Confer et al. A lot of common questions about evolutionary psychology are answered in the paper.
• The Social Brain Hypothesis by Robin Dunbar. An example how evolutionary hypotheses can be tested.
• Human Nature and Culture by David Buss. Culture differences often stem from different environments.
• Where Evolutionary Psychology meets Cognitive Neuroscience: A précis to Evolutionary Cognitive Neuroscience by Krill et al. Three-word scientific areas are the future: evolutionary cognitive neuroscience already arrived. The paper contains a lot of interesting facts about our brains.
• The Presence of an Attractive Woman Elevates Testosterone and Physical Risk Taking in Young Men by Ronay and Hippel. Another example of how evolutionary theories can be tested.
• The Commitment Function of Angry Facial Expressions by Reed, DeScioli and Pinker. Sometimes evolutionary psychology is almost pure game theory.
• The psycho gene by Philip Hunter. There is growing evidence that there is a gene for psychopathy.
• Feminism and Evolutionary Psychology by Buss and Schmitt. Feminism is one of the ideologies most affected by evolutionary psychology, but the authors also note that evolutionary theories predict more similarities than differences between genders.
• I often go to www.epjournal.net, an open access academic journal, meaning that the articles are accessible for free. It's much more useful than reading the news or mindless browsing the internet.