> That mammalian brains are motivated by optimization of energy expenditure. Like anything in biologically evolved systems, energy efficiency is key, since food is often scarce.
That doesn't square with empirical reality. Evolved biological systems appear to be optimized for robustness to perturbations, not efficiency (John Doyle argues that there is in fact a fundamental tradeoff between robustness and efficiency, for all types of complex systems not just biological).
> how do mammalian brains problem solve using just thoughts.
They don't. Sensory input is required for brains to learn new classes of problems.
> find a way to train RNNs that are able to adapt to new problems
Is this something different than multi-task learning?