Ramanujan had a phenomenal ability in mathematical logic. Shakuntala Devi had the ability to perform mind-boggling calculations at a breath-taking pace.We can learn many things from knowing that humans like Ramanujan and Shakuntala existed. One is that there is a configuration of our brain that can result in such super-human abilities. Both Ramanujan and Shakuntala lived pretty normal life otherwise. This tells us is that a normal brain is not at a Pareto optimal level. Which is good news because it means that one does not have to sacrifice any mental functionality to achieve the abilities of Ramanujan or Shakuntala Devi. Another observation - they were both born in pre-independent India in poor families. Ramanujan received his education nothing out of the ordinary. Shakuntala Devi was not formally educated in school. This implies that starting your kids early in the tried-and-tested mechanisms of teaching so that they are successful in school early in their lives may not be optimal. It is low-risk strategy which will dumb down kids.
Brain is a physical object after-all. And, it operates by sending signals between its millions of neurons. This implies that there is a certain connectivity and configuration of neutrons that can allow one to think like Ramanujan or do calculations like Shakuntala. Can we design such a brain? Not at present, but maybe in the future. What do we need? We need to first figure out what is the critical size of the configuration of neurons that needs to be trained. Ideally, one would like to tailor the entire brain, but that may not be practically feasible. Once we know the critical size of connections that we need to optimize, then we need to figure out what to optimize. Which configuration results in a Ramanujan brain? Is that configuration unique? It is most likely not unique, because, first, in that case the probability of existence of Ramanujan becomes diminishingly small. Secondly, that would mean that any new experience or learning would change this configuration and the brain will become sub-optimal again. So, an ensemble of configurations should work. Great! But, what is the objective function that should be optimized? I do not think we know the answer of this one as well. Once we have figured out these questions, then it comes to the logistic part of physically designing such a configuration. How do we assemble neurons in a configuration? We will probably not start attaching single neuron to others from scratch. That's not possible. We will probably need to scan thousands of brains for their connectivity and select candidates that have closely related networks. This exercise will also tell us if the brain's ability correlates positively as we move towards the optimal configuration in a continuous manner, or will we see a step-change in the intelligence. If the process of optimization of networks is invasive, one needs to somehow grow that brain in-vitro. Then, we start optimizing the connections.
