Watson, Richard A., and Eörs Szathmáry. “How Can Evolution Learn?–A Reply to Responses.” Trends in ecology & evolution 31, no. 12 (2016): 896-898.
URL1
The commentaries [1, 2, 3, 4] on our target article [[5]https://www.cell.com/trends/ecology-evolution/fulltext/S0169-5347(16)30180-X?rss=yes&code=cell-site#bib0025)] raise important points that warrant some discussion and clarification. Blute [1] is correct that there are many aspects of evolutionary biology that correspond to different forms of learning and that we did not address all of them in our Opinion paper [5]. Where we focussed on the evolution of developmental organisation, ecological organisation, and transitions in individuality, Blute emphasises a different connection. The evolution of environmentally-sensitive developmental plasticity is also a type of learning. In general terms, learning systems are characterised as systems with multiple timescales of adaptive behaviour where the parameters controlling an ‘inner’ adaptive loop are actually variables that are modified by an ‘outer’ adaptive loop (Figure 1C). In conventional learning systems both loops operate on within-lifetime timescales, in other words the knowledge, biases, and proclivities that shape short-term behaviour (inner loop) are not fixed and are instead subject to the long-term experience (outer loop) of an individual. In the evolution of evolvability both loops are shifted into multigenerational timescales – in other words the constraints and biases that shape the path of micro-evolutionary trajectories are not fixed but are themselves modified by natural selection (Figure 1A) [5]. The evolution of environmentally-sensitive developmental plasticity (Figure 1B) addresses an intermediate timescale where the outer loop is evolutionary (natural selection over multiple generations) and the targets of that outer adaptive loop are the parameters of an inner adaptive loop (or reaction norm) controlling how phenotypes respond to environmental cues and conditions within lifetimes.