I love that I’m safe. I love that I’m in a microcosm of a world that’s allowed to exist as a microcosm, that I have food, a home, a bike—that everything’s so easy.
I love that I’m healthy. That my body always works with me, and that the voice in my head is cheerful and curious a great majority of the time.
I love that I have family and friends who love me as close to unconditionally as should ever be warranted. I am grateful for the people surrounding me, observing with bright engagement and welcoming eyes.
I love that I feel empowered. That I feel like my thoughts matter, that I can make a difference in the world, that people will listen to me and I can accomplish something significant with the effort that I put forth. It’s so improbable to be one of those people whose names I admire now. But through a series of continuous effort from people around me and in the past, a huge helping of luck, and trying on my own, I’ve realized that it can happen. That I have the ridiculous chance to be one of these people who change things in the future. These people who are not impossible unmovable figures, but could be someone who has gone through these events in her life, who has made these sets of choices, who has had these sets of interactions, who can end up where I have planned. It’s humbling; it’s incredible. That what I do might matter.
(And doesn’t every experience contribute to that life goal, to a greater or lesser extent? So many skills that need to be acquired, hard and soft; so much personal development that needs to happen, to be the best person to portray knowledge. Self-care, to keep everything functioning (everyone works better happy). So many choices, some good, some bad; useful experiences where you don’t expect them; myriad useless experiences and lacking situations. That’s luck and skill in avoiding them; bravery and the principle of exploration in trying new things anyway. I suspect this is the clearest description of my life philosophy that I’ve elaborated.)
I love that I live in a time and a space where what I want to study exists. I love that scientists are no longer just researchers: that there’s expected to engage with the outside world, manage apprentices, write, and read. I love that there’s funding for us to study, acceptance and respect within this job, (that our society functions and embraces science and there’s a slot if you follow the well-worn path.)
I love that we’re in an age of information! That the Internet exists! That people write books, that the knowledge is out there, that people are interested, that there’s so much to know. That if I want to take a class on Natural Language Processing, there’s a free course online that I can sign up for; that there’s lectures on Youtube on anything, that it’s so easy to pick up a book in the bookstore that will change the way I think. That if I want to know more it’s free, that I can make the time to do so, that I can talk to people and learn about ideas and there’s just so much knowledge.
Prof. Tom Griffiths wrote this paper on priors. It starts out like this: you show people mathematical relationships like the linear equation y = x, negative linear equation y = -x, random dots, or a parabola. These relationships are presented in a piecemeal fashion so that you’re presented with individual (x, y) pairs– you never get the whole picture. You then take the training data away and have people try to replicate these relationships in the same piecemeal way—trying to guess that the y value is for a given x.
You then pass the guesses from the first person to a second person, using the first person’s guesses as the second person’s training data. And continue doing this for seven more people. What kind of mathematical relationships emerge at the end?
Answer: linear relationships. Sometimes negative linear relationships, but almost always linear relationships. No matter what mathematical function you start out with, you end up with a linear function. How strange is that?
The concept is that people have these things called “priors” in their heads—intuitive ideas of how a value x and a value y should be related. And when you’re communicating this type of relationship to other people in a piecemeal way— when you pick up information as you go along, never seeing the big picture—people return to their “priors” rather than the “evidence” that they’re seeing.
I’m fascinated by the idea that people have these priors on a large scale, not just for the low-level example of mathematical functions. That people have intuitive ideas they return to as societies and in history—that these are the things that fundamentally make us human, that we can overcome with data perhaps, but that we return to. It’s a framework I only thought of last week, but it’s been immensely satisfying. I’ve been waiting to see if it’s a life-change moment: it’s still too early to tell, but it’s very much influencing my actions this week and last. All of sudden, so many things are interesting; there are so many things that could fit in within this framework that I have this new burning need to learn.
I know a few people who are innately riveted by certain topics, and do very well at those topics and don’t care enough to do well in others. I work differently: I’ll be interested in any topic that I convince myself I need to be interested in. (And ignore everything else: I have a strong spotlight of attention.) This is a topic I am very, very interested in, and it encapsulates a whole heck of a lot of other topics, so I’m as pleased as I can be. I’ve never seen myself this curious before about such a range of things. And it’s such a tiny switch—just a slightly different generalization of ideas I’m already well aware of, and suddenly all the things I’ve been ignoring have been flagged for inspection. It’s making me more curious about the world—it’s motivating me to seek broader learning out in a way I’ve never seen myself act before. In this moment I have the time, I have the mental energy, and I now finally have the motivation. I suspect it’s my most innocuous life-change moment—a Formal Hall conversation with Stephanie and Vasilis which spun my thoughts down numerous paths.
And– this is the kicker—it’s related to what I want to do in the future. I want to study in the field of computational cognitive (neuro)science. A phrase that’s batted around a lot in the field right now is “Bayesian models of cognition”. To break it down by word:
“Bayesian” refers to “Bayesian statistics”, one of the two most common types of statistics. You probably know “frequentist statistics”, though not by that name—it’s the stats you learn in high school, where you test hypotheses by determining if one is “significantly different” compared to the other. Bayesian statistics is another way of representing how confident one should be one idea compared to another, just like frequentist statistics. However, Bayesian statistics is more relative—instead of saying, “this idea is better than that idea”, you say, “I have this degree of belief in this idea, and this degree of belief in this other idea, based on my priors and the likelihood of the evidence you’ve presented to me.” If it means anything to you, Bayesian statistics deal with whole probability distributions rather than just probabilities. Bayesian statistics were popular in the mid-1900s but fell out of use due to lack of processing power in our computers; it’s recently become popular again. A lot of the artificial intelligence breakthroughs these days are based on Bayesian methods.
“Models” refers to computational models—theoretical and mathematical descriptions of how people behave as reflected through behavioral data. “Cognition” refers to how people think—there are more precise definitions online, but I think of cognition as understanding the processes by which we understand the information coming into our senses.
When I go on graduate school visits, or talk to any scientists who asks what I’m interested in doing in the future, the phrase “Bayesian models of cognition” is specific and technical enough to satisfy most inquiries. The paper I described earlier by Prof. Tom Griffiths falls squarely in this category. And if you’ve forced your way through the last few paragraphs you hopefully have a grasp of what I’m talking about as well: I’m interested in understanding how the mind works, by formalizing its processes in mathematical models that will allow us to predict and explain behavior. Behavior which is explained in the Bayesian framework: i.e., by balancing priors—these intuitive ideas we have in our heads—with likelihoods—the evidence that we’re presented with.
Of course, there’s a huge difference between the kind of work I might be doing in graduate school with Bayesian models of cognition, which are usually small-scale, and trying to figure out what the underlying priors and likelihoods of the entire human race are. I don’t actually think that would be achievable in my lifetime, not that I have any basis for making that assumption: it just seems like the amount of categorization and evidence and formulation and ideas you need to assimilate is ridiculously large. People have been trying to do it since philosophy was born, as well, and I don’t know if the additional knowledge and frameworks we’ve assembled at this time would allow me to make any useful progress. I don’t even know if I’ll stick to this as a life goal: it’s quite likely I’ll lose interest or I’ll figure out it’s too big and scale down or realize the whole idea is hopelessly naïve.
But for now, it’s making me want to look up lectures and read and discover things; making me me curious and seek out learning, and I can’t really be happier than that.
Happy Easter readers :).