According to Miller, the failure of the mating scenario may reflect our prehistoric ancestors not realizing that mating could result in children because of the nine months between mating and birth.
This seems to imply that successful evolutionary neural behaviors are consciously learned, as opposed to being "learned" over generations due to natural selection [0], in a similar way to how a neural network is trained.
[0] https://academic.oup.com/bioscience/article/50/10/883/234025 William G. Wright; Neuronal and Behavioral Plasticity in Evolution: Experiments in a Model Lineage: Evolutionary changes in sensory neurons correlate with changes in learning phenotypes, BioScience, Volume 50, Issue 10, 1 October 2000, Pages 883–894