Congratulations on making it this far! Hopefully you've enjoyed this quick survey. You're approximately up to the state of the art in both neuroscience and machine learning, however both fields are changing very quickly because they're informing each other, and the rest is up to you.

If you're in this field already, all I can say is "math math math", and if you're thinking of studying this field, I can recommend a good foundation in biochemistry and cellular biology, as well as the physical sciences (electronics is very good, and any day now there will be commercially available photonic chips).

Neural systems are complex because they're nonlinear and they can exhibit all manner of structured and unstructured dynamics. A big part of the difficulty in modeling such complex systems has been visualization, and secondarily number-crunching, and AI is helping in both areas. However there is no substitute for learning the basics, and if you're into neuroscience that includes most of machine learning, and vice versa. At this moment the transformer design is informing us about the requirements for the cingulate cortex, and vice versa. Higher areas like the frontal cortex are very difficult to study if we don't know what to look for. The timeline architecture presented on these pages assists us when considering time scales. How is it that we can get an abstract motor sequence to play out along the timeline in a precise manner? Conversely, how can we store a complex motor sequence efficiently so it can be recalled on demand and played back at any speed?

The placement of the hippocampus at the back of the timeline is obviously of great significance, and only partly because of the curious phase encoding mechanism used to reduce the dimensionality of complex scenes. The mechanism of contextual retrieval through the frontal lobe is currently of extreme interest, and many groups are designing experiments to tease apart the components of this complex interaction.