After a barren few years for popular science books, I stumbled onto several good ones this year.

I first encountered evolutionary psychology about a decade ago in its former incarnation as social biology but was rather put off by it then, when it was rather short on data and long on theorizing, inevitably becoming a mirror for the writer's political ideology. The field of evolutionary psychology has since progressed in leaps and bounds, and one of the glittering lights of the field is Jonathan Haidt, who also happens to be a gifted writer. His "The Righteous Mind" outlines a persuasive case that there are several underlying psychological modules that define human moral thinking, which to rattle them off carelessly are: care/harm, fairness/cheating, loyalty/betrayal, authority/subversion, liberty/oppression and the most interesting to me, sanctity/degradation. This theory is backed by oodles of cross-cultural research and steeped in careful evolutionary thinking. I think combining these insights with the work in evolutionary religion (David Sloan Wilson) will provide a doorway into 21st century religion.

As a scientist working in biology, looking for a connection to medicine is an important skill to improving one's grant-baiting ability. Sadly, this does not mean I actually know anything about medicine; my knowledge of the history of medicine is rather meagre. I fortunately got a chance to fix this when I stumbled onto James Le Fanu's "The Rise and Fall of Modern Medicine", filling a rather surprising gap in the literature. It does this through a rather provocative hypothesis – that there was a golden age of medicine, and it has already passed. The first half of the book provides an extremely readable history of modern medicine, mainly over the course of the 20th century, structured in terms of 11 pivotal breakthoughs. This half of the book is wonderful and has definitely enriched my understanding of modern biology. The second half, where Le Fanu rails against modern day medical research, is rather uneven, and can be safely skipped.

In biology, two of hottest areas are neuroscience and developmental biology. In Rob DeSalle and Ian Tattersall's "The Brain" we have a exposition that beautifully syntheses the two. This book explores,not just the brain, but the whole notion of a nervous system in terms of developmental biology. To do this the book reaches right back into the tree of life to trace the evolution of organic life, and at each stage of development, feeling out the functional requirements of a working nervous system, culminating into the nervous system of the human brain. Unlike other lesser books on the brain, the brain doesn't even make an entry until 2/3 of the way into the book. More that the generally breezy quality of the writing, what impressed me most was how clued these guys were into the cutting edge of biology – they write about the fascinating world of non-coding RNA, transposonic elements, and the even more mysterious process of RNA editing. These are findings I've found to be still outside the mainstream (see for example many bioinformatician's resistance to the idea that transposonic elements have any biological function at all).

I have had a hole in my technical education, ever since I skipped second-year Linear Algebra for what I thought would be the sexier subeject of Introductory Philosophy. It's been a problem ever since as standard linear algebra theorems are often invoked to explain higher level physics. I've tried to rectify this over the years, but each time I'd try a new linear algebra text book, I died a little. The difficulty with teaching Linear Algebra is that there are no really overarching equations that binds it all together, instead there is a forest of numerical techniques to master, as well as a bunch of disconnected abstract algebraic theorems. As a consequence, half the linear algebra textbooks soar into abstractions which obscures the messiness of practical techniques, whilst the other textbooks dive into the nitty-gritty algorithms, but get lost in the ability to make sense of all the different techniques. Gilbert Strang's "Linear Algebra" is that rare Linear Algebra textbook which perfectly straddles the two. It got me interested to learn all the theorems, which were presented in a way that handily motivates all the key techniques. You will learn how to reduce matrices into various canonical forms, but also how to use the abstract theorems to grasp the meaning of the solutions. As a result, I can now plow through some machine-learning papers in bioinformatics that has had me stumped for a while.

This was also the year that after many aborted attempts, I finally got my head around information theory. I settled on David MacKay's "Information Theory, Inference and Learning Algorithms" because it was the only one that was willing to motivate all the theorems of information theory for a rank beginner. Before every theorem, MacKay patiently describes in elaborate and necessary detail the kind of problems each theorem would solve. Information Theory can be a beautifully abstract theory, but its power is in its ability to illuminate very specific computation problems. I loved the focus on practical consequences such as the explanation of how the gzip algorithm works as an elaboration of the use of entropy in encoding files. But more importantly, MacKay has taken the idiosyncratic (but ultimately correct) approach of teaching information theory as a branch of bayesian inference. The information theory then becomes a launching pad into machine learning. Thanks to this, I also got to learn the basics of machine learning for free. Booyah.