A new factor in evolution

We invoke the Baldwin effect in dalla2026between to clarify the scope of within-lifetime learning: developmental optimisation of an encoding does not replace evolutionary convergence, it can accelerate it by raising reproductive success of organisms whose architecture supports such learning.

Baldwin's original proposal that learned behaviour can shape selective pressures motivates how we read the interaction between token-level training and lineage-level evaluation in riva2026task, where post-training injects evaluation signal into the next-token objective.