This study may hold some keys to advancing HTM. It makes me wonder if all cortical columns have reinforcement learning capacity. Incorporating this reinforcement association mechanism may be another step towards AGI.
Just like the coffee cup example where the predictions simultaneously happen in all directions, reinforcement learning may happen throughout the “thousand brains.” Consider our capacities as humans, we can learn to find reward in the most abstract notions all the way down to simple sensory experiences.
Best regards,
Jack