I see a skill as a specification for action, rather like an instruction manual or a computer program. I see the skill as providing general instructions on how to act, much like the text and diagrams of an instruction manual. I focus on physical actions that involve motion of the skeleton by means of joint rotations. The skill organizes the action by specifying the order of movements and their coordination, without specifying the precise timing or the specific forces applied to the muscles. It provides general instructions of what to look for and how to make adjustments.
An application of the skill translates the general and abstract instructions into action by generating specific control commands that tell the muscles just when and how much to rotate the joints. It specifies where to point the head and the eyes to get visual information and how to use it to select or adjust control commands.
From the perspective of evolution, a skill is a trait. Like other traits, skills evolve over generations and drift, bifurcate, etc. I see an innate skill as encoded in the DNA and also stored in the brain.
From the perspective of information processing, evolution gives us two views of individuals within a species. One is as DNA, and the other is as living, functioning creature that survives and reproduces. From an information content perspective, the two views are equivalent. DNA is a blueprint or specification of the individual from which the complete individual can be constructed. In most cases, this DNA blueprint is all that the individual passes to the next generation. Evolution 'improves' a species over many generations, it does not 'improve' or benefit the single individual. In other words, the information relative to a single individual is conserved.
I am particularly interested in the skill and action component.
I see evolution as changing the information encoded in the DNA, whether for the skeleton or an innate skill. I see this change as intergenerational, where the rate of change and the direction of change are of particular interest. Change presumable starts with single individuals and eventually affects population averages. We can see that this change might take very many generations until the offspring of the mutated individual 'take over'.
To get to the optimum value of the attribute will take even longer, since the population distribution may have to shift before we can expect a mutation out along the tail as far as y, as shown on the graph.