The main question is how much the combination of experimenting, imitating, and learning can speed up the discovery and diffusion of new, more beneficial skills.
I see experimentation as analogous to evolution. New experimental alternatives have to be generated, somewhat like mutations. Many of the experiments won't work and may even be fatal. If an experiment leads to a desirable outcome, that particular design and treatment are likely to be repeated. This evaluation component is similar to the fitness measure for evolution.
I see imitation as an alternative form of propagation. Both propinquity and perceived success are important, somewhat like mating rituals.
Learning is a necessary component for both experimentation and imitation, to be able to add skills to the skill set. The combination of experimentation, imitation, and learning therefore function to 'evolve' skills through a second route that complements the evolution through DNA.
I have started to work on models of such skill evolution, with skill drift and diffusion, exploring different assumptions