Feedback

I’ve been thinking about feedback and feedforward loops as I read about the origin of life and the emergence of dynamical control systems to optimize robust behavior. Much of it is abstract, and sometimes I get lost in the math. I consider myself a beginner in these areas, and I’m experiencing the challenge and frustration in my lack of ability to teach myself efficiently. It’s taking time and concentration that I can no longer spare as teaching and the new semester begins. Now if only I had a good teacher that can help guide me along the path, helping to predigest some of the material and providing feedback so I don’t go around in circles.

 

All this is, however, an excellent reminder of why feedback is important to learners. And while the basics of college chemistry is old hat for me, it is new to most of my students. In the same way that I feel like I’m fumbling around systems-theory, my students will likely be fumbling around thermodynamics, kinetics, and equilibria. My job as a teacher to help their learning is to predigest and sequence the material in a way that aids the student to begin to get a grasp of the material. I say “begin to get a grasp” because learning new things (you are not evolutionarily pre-adapted to learn) is an iterative process. You can’t magically understand this knowledge without the work and effort. So, yes, the learner needs to contribute substantially to the effort. And it helps to get regular useful feedback along the way, to help ease the difficult journey.

 

In days of yore, constant feedback was provided one-on-one through apprenticeships over years – the artisanal model. There was no other way to learn expertise in such skills, at least to a high-enough level to make a living as a specialist of some sort. Nowadays, education is mass produced. This isn’t necessarily a bad thing – many basics of the basics we learn in school can be taught reasonably in a classroom with many students. Hopefully not too many, because feedback, and individual feedback in particular, takes time and energy. Let me put it this way: There are many commonalities in how students learn (which lend themselves to mass methods) but when a student isn’t “getting it”, this failure comes in a variety of forms less common to each other. It’s somewhat like the adage that “all happy families are alike, but every unhappy family is unhappy in its own way”.

 

Class sizes have slowly been creeping up, at least in my department at my college because student numbers are increasing but faculty numbers not so much. Not only do I know my students less well on average in a larger class, the time required for optimal individual feedback scales with class size. (Not to mention, a faculty member will feel increasing time pressure with higher administrative burdens and research expectations as the years pass.) What has happened in my general chemistry classes? My colleagues and I made the transition to “smart” online homework systems that are packaged with any standard G-Chem textbook. The system gives them hints and feedback if they get an answer wrong and tries to nudge them along the path. A provider might make claims about how smart the system is in individual targeted-ness, although this is mostly a mirage.

 

I’m not all that happy with this state of affairs, which is why when I get to teach smaller G-Chem classes (e.g. Honors), I write my own problem sets in addition to the online homework. In office hours, most of the questions I get are on the problem sets (rather than the online homework), and I get to engage in feedback when answering questions, and when I eventually grade them. These conversations allow me to get to know individual student quirks and who might stumble how and where. But I don’t do this for my larger G-Chem sections. Call it laziness. Call it prioritizing. I’m not satisfied with this situation, and I have tried to mitigate that loss with frequent low-stakes quizzes (with in-class immediate feedback) and other measures.

 

The challenge with computer-based feedback is that it ultimately relies on algorithms honed through mass production. This means that it’s good for a subset of things you’d want your students to learn to do, but not everything, and possibly not even the lion’s share. I recognize that with advanced parsing techniques and machine-learning methods, algorithms will make more headway and be able to adequately cover a larger subset, but there may still be much that’s missing. I do some Duolingo every day, and I can see how the techniques they use leverage the science of learning, I would likely still not do great if you plopped me into an environment solely with native speakers. I suspect there’s a fundamental limitation to computer-based approaches because complex systems are likely non-simulable, non-computable.

 

I don’t know the solution out of education’s iron triangle. But one thing I do know is that if I want to help my students learn, I need to spend the time and energy giving them good feedback. I have a suspicion that over the years I have lagged in that area as other things started to make demands on my time. I suppose I’m writing this to remind myself of the importance of giving timely and sufficiently detailed feedback.