My first patent was published today - super proud of the work @clayton_t22 and I have done to get here. Leading the way @DrivelineBB

Keep it moving

Neural circuits adapt via biological gradient descent, and movements are refined via Bayesian sensorimotor updating. Both processes happen together, with the brain constantly correcting and predicting to learn something new

As the underlying synapses are still changing, the same parts of the brain are used to make movements become even smoother and more accurate > motor cortex adjusts plans > cerebellum fine-tunes corrections > spinal cord integrates sensory updates

The main idea is that Posterior (updated movement plan) = Likelihood (sensory feedback) × Prior (expected outcome). This runs in parallel with gradient-based learning, which is conceptually fairly similar to Bayesian optimization

Alongside the synaptic changes that occur during early motor learning, the brain also predicts and updates movements by combining prior knowledge with sensory feedback, a Bayesian principle

Early motor learning is about controlling forces and minimizing errors. Novices rely heavily on feedback, making their movements slow and inconsistent. But, trial and error helps shape new neural circuits over time

There's a lot going on in the brain at this point > motor cortex forms new motor plans > cerebellum corrects errors > basal ganglia runs a dopamine-based feedback loop > spinal cord reflexes start adapting New synapses are also forming via synaptogenesis, neural firing

Early motor learning is largely your brain running a noisy gradient descent algorithm. It's all about using trial, error, and feedback to minimize movement errors

tbh I don’t think we tell kids how small the window is to be exceptional at many physical or intense intellectual pursuits. You basically have from 15-35 to become meaningfully world class at many things, which usually means starting in early teens

That isn’t to say you shouldn’t get technical. There’s a time and place for detailed, well-formatted info, just don’t superdome an athlete by dropping it all on the title page

The more I work on these types of projects, the more I come back to the basics > color coding for intuitive associations > bars and scores for easy interpretation > histograms for population context > thoughtful use of negative space and text

Spent the past several weeks rebuilding @DrivelineBB composite score models and reports. When making deliverables like this, it’s all about keeping the end user in mind For anything built around an assessment, you should be able to instantly answer “How good am I?” or "Did I get

Get your hand-eye work in

We have officially invented something @bl0ebaum @DrivelineBB. 💻👀📈

+1 for the good guys

obviously

A man poses in front of the Soyuz rocket monument, Baikonur, Kazakh SSR, 1980s

Meeting of ages. Osaka, 1960