Allen Wang

Notes on Human Qualities

Parent article: Lessons from Undergrad

During the earlier years of undergrad, I thought very hard about what kind of person I wanted to be. In the end I ended up becoming the person I always was - with a bit more maturity. But in that process, I gained better understanding of some human qualities. Here are blocks of thoughts from late nights/in the shower during the past few years.


When I was a kid, I thought leadership was some magical power. It sounds absurd, but anything we don’t understand tends to seem magical. In actuality, leadership is a much more tangible concept.

A leader is someone that others want to follow. But who do we want to follow? Logically, people follow whoever that benefits them by following. Emotionally, people follow whoever that comforts them. It’s someone that can show people answer when they’re are lost, unite them when they’re divided, or give hope when they’re scared. In the caveman’s time, the leader was someone who brought back food so the tribe could survive. In business, the leader is someone that can figure out how the company can make money. In politics, the leader is someone that can help build a better nation. During the OS labs, the leader was someone that could fix memory leak at 4am so we could go home.

So it comes down to having a direction, the ability to communicate it, and skills to execute. In most cases, good leaders also deeply care about others - people naturally want to follow someone that cares for them. But someone chasing a selfish goal can also end up inspiring others.


Courage is often confused with fearlessness. But it’s is not the absense of fear; it’s the ability to overcome it.

** As I’m writing, I found out that there are famous quotes with almost the exact same wording (I didn’t steal it!). Nethertheless I think it’s important to elaborate what it means. **

Fear is a heuristic meant to keep us from danger. It’s largely processed by a structure called the limbic system in the inner part of the brain[1], which developed early in evolution. To be fearless would mean that this primitive structure is less sensitive, and that could be convinient; it leads to more risk-taking behaviours and pontentially higher rewards.

But fearlessness should not be the virtue to seek for. First, we can’t always control whether we feel fear, because it’s a physiological response. Second - a somewhat philosophical point - to pray for fearlessness would mean to count on the mere chance that the emotional response is absent; and to surrender when it’s present. In other words: if we don’t feel fear, that’s great - but if we do feel it? Do we just accept that we’re not fearless? Certainly, that doesn’t sound very courageous.[2]

Courage means facing and beating the fear. It involves justifying our purposes and suppressing the internal voices that don’t cooperate. It’s a very rational act; it means assessing our emotional reponse and harnessing it with reason.


Somewhere along the way of trying to become smarter, I got interested in intelligence itself. It’s what I think about a lot these days, and there are a few lessons I got from studying cognitive science and machine learning.

Intellectual Relativism

Intelligence is essentially a form of information processing. Sensory inputs such as vision and sound are converted to neural representations, which are firing patterns of the neurons, and the brain forms ideas from these patterns. What we call problem solving happens in the firing of neurons, where the input patterns are transformed to patterns corresponding to the answers. Though people disagree on the details, that’s roughly how the brain works.

This understanding gives a broad view to intelligence. Many people think that being “smart” is synonymous with having strong analytical skills. Some even think that math and theoretical physics are the highest forms of human intellect. But underneath the skull, the thoughts for math and physics are simply firing patterns of the neurons, and not inheritantly more complex than other cognitive activities like composing music, understanding emotions, or playing soccer. In fact, playing soccer probably requires more brain power than solving math problems. It involves keeping track of the behaviours of 21 other humans, coordinating hundreads of muscles in the body, and making strategic judgements, all at the same time. To do so the brain needs to simulate all kinds of dynamic models of the world - most of which are done unconsciouly[3]. Mathematical theorems, on the other hand, can sometimes involve just few objects and properties (if you don’t believe this, you should try learning abstract algebra!)

But while many people can play soccer, fewer can understand advanced mathematics. It’s because the act of playing soccer recycles the abundant brain structures dedicated to scene understanding, body coordination and high-level decision making, which has taken shape over the evolution of mammals. On the other hand, mathematics requires abstract and rigorous thinking, which the brain is less wired for. As a computation, mathematical thinking may not have higher complexity - it just takes a different neural structure that needs to be nurtured. So there isn’t one form of intelligence that’s inheritantly supeperior to others. Some just happen to be more scarse, and some are more useful in an given era.

These ideas motivate me when learning new skills, be it a sport or a rigorous topic - “I’m training a new structre in my brain.” They also humble me to appreciate all forms of human intellect.

Learning, Fast and Slow

The ability to learn things quickly is often appraised, but less so is the ability to learn things really well. They are not the same; there are cases where people become functional with new concepts very fast but don’t end up fully grasping them, and other cases where people need more time digesting ideas but eventually develop a deep understanding.

Learning a new concept involves relating it to existing knowledge base. Knowledge can be modeled as a graph; a concept is a node, and is understood once it’s connected to enough other nodes. This, for example, explains why it’s easier to learn new ideas in a domain you’re already expert on than to grasp ideas from an unfamiliar domain; it’s to easier build connections to exisiting nodes nearby than to develop a new subgraph. To understand things well means to develop many connections between ideas, so that the knowledge graph is dense. Doing so inevitably takes more more, but once done, allows you to retreive and connect ideas more easily.

We can observe a similiar principle of learning fast vs. slow in machine learning algorithms. Support Vector Machine (SVM) is often the go-to algorithm for small datasets, but is usually outperformed by deep neural networks when more data is available. First, the search space for optimal parameters in SVM is simple (i.e. convex), making the optimization problem easy. With a smaller hypothesis space, it’s also less prone to overfitting even with little data - this is illusrated by a rigorous result from statistical learning theory[4]. But the small hypothesis space limits the accuracy of SVM on many problems. Deep neural networks, while harder to train and require more data, can approximate more complex functions because it has a richer hypothesis space. It’s a relatively slow learning algorithm that can learn more deeply.

We might feel frustrated when when we can’t learn things quickly enough. But these ideas should suggest that taking more time to understand concepts is sometimes necessary. After all, we are all some forms of deep neural networks.


As we become older, we tend to grow too comfortable with who we are; we get into respectable schools, get respectable jobs and start receiving some recognition. We establish our place in the world, get into a flow, and life slowly becomes flat.

The common wisdom is to find peace with reality, and it’s considered a part of becoming an adult. I’ve learned to do it to some extent, and I must admit; it can feel pretty good. But I strongly suspect that the wisdom is simply a defense mechanism so we could avoid pain. As a child, I was driven by a stubborn resistence to reality. My stubbornness brought me extra pain, but I think it also gave me the escape velocity to reach a bit further than I would have otherwise.

What I’ve learned, however, is that I need a body that can withstand the enthusiasm. Eating healthy, exercising, and getting enough sleep are all part of that. My mother once told me that life was a marathon. This, to date, remains one of the wisest advices I’ve received. I intend on running that race till the end.


Sadly, I don’t know enough about this one yet. But I think that this might be the most important quality. I’ll get back on this.

  1. It’s thought to be processed by the amygala, though not entirely
  2. This is all ignoring whether or not we have free will. But let’s not get too philosophical :)
  3. Another interesting fact is that the brain consumes roughly the same amount of energy no matter what you’re doing - because the basic task of keeping the body running takes most of your brain power (literally)
  4. The Fundamental Theorem of Statistical Learning

comments powered by Disqus