In the mid 1990s, Günter Wagner[1] argued that modularity could have arisen and been maintained through the interaction of four evolutionary modes of action: [1] Selection for the rate of adaptation: If different complexes evolve at different rates, then those evolving more quickly reach fixation in a population faster than other complexes.
[2] Constructional selection: When a gene exists in many duplicated copies, it may be maintained because of the many connections it has (also termed pleiotropy).
[4] Compounded effect of stabilizing and directional selection: This is the explanation seemingly favored by Wagner and his contemporaries as it provides a model through which modularity is constricted, but still able to unidirectionally explore different evolutionary outcomes.
The semi-antagonistic relationship is best illustrated using the corridor model, whereby stabilizing selection forms barriers in phenotype space that only allow the system to move towards the optimum along a single path.
This allows directional selection to act and inch the system closer to optimum through this evolutionary corridor.