I'm not quite sure I understand what you mean, so let me frame my view on labor from the academic economic perspective:
Labor ( and thus labor efficiency ), is always assumed to be an equal portion to the output give, multiplied by human capital H. So our production function is equal to a cobb douglas production function Y= f(k)=A*k^alpha*h*L^(1-alpha). Qualifiers for efficiency are measured by our expanded cobb-douglass, where A ( productivity ) is measured by TxE ( a function of technology and efficiency). Technology is a function of the portion of population involved in R&D, over the cost of the project ( in terms of labor), multiplied by the total labor force. This means that increases in R+D contribute to increases in efficiency of the capital. Likewise, efficiency (E) is a factor of market influence ( proximity to outputs, business relationships between input output firms, etc). On our labor side, h ( human capital ) is measured in terms of number of years in education, where each marginal year gives an increase in the value of human capital. If you decide to use the cobb douglas on a per worker term, you divide the entire function by L, so your CD function becomes y=A*K^(alpha)*h^(1-alpha).
So it is in the best interest of the corporate world to have both max efficiency from the work-base, as well as for them to be educated. A unit of input is modified by its human capital premium, where the base unit of input ( in your analogy of a calorie ) is the base unit of input multiplied by h.
