Phenotypes to remember: Evolutionary developmental memory capacity and robustness

While the theory of neural networks can (and does) infer the same conclusions based on different representations, in the case of modelling real biological situations the adequacy of the representation can be crucial (the same holds for neuronal networks). Our results show that a linear change to the representation has profound impact on the essential features of the system. While in the customary (neural) {-1,+1} representation there are no neutral elements in the interaction matrix, the biologically adequate {0,+1} representation of genetic regulatory networks allows for the free choice of interaction elements being opposite to “0”. This feature turns out to increase the robustness of the system against the disturbance of interaction coefficients only if the system is very sparse, which guarantees the commonly zero elements in the embryo and adult states. Another feature of our representation is the large number of different interaction matrices entailing the same developmental process, thus evolution “from scratch” does not face so many constraints. In other words, starting with a single ancestor an extraordinarily rapid morphological diversification could be attained, which is the hallmark of adaptive radiations.

The authors thank Ferenc Jordán and Harold. P. de Vladar for comments on an earlier draft.

All relevant data are within the manuscript and its Supporting Information files. The source code of the implementation of the model is available on GitHub (https://github.com/peterszabo77/developmental_memory_2020).

This work was supported by the National Research, Development and Innovation Office (NKFIH, https://nkfih.gov.hu) under OTKA grant numbers K124438 (A.S.), K119347 (E.S. and A.S.) and GINOP-2.3.2-15-2016-00057 (E.S., P.S. and A.S.) research grants; the ATTRACT Project EmLife (AS, ES and MS) (https://attract-eu.com); The Volkswagen Foundation (https://www.volkswagenstiftung.de) (initiative “Leben? –Ein neuer Blick der Naturwissenschaften auf die grundlegenden Prinzipien des Lebens”, project “A unified model of recombination in life”) (A.S., E.S. and M.S.). A.S. was supported by the Bolyai János Research Fellowship of the Hungarian Academy of Sciences (https://mta.hu/) and the ÚNKP-19-4 New National Excellence Program of the Ministry of Human Capacities. M.S. was supported by the research grant CGL2017-89160-P from the Ministerio de Economía, Industria y Competitividad, research grant 2017SGR 01379 from Generalitat de Catalunya, and the MTA Distinguished Guest Fellowship Programme in Hungary. All authors were supported by Templeton Foundation under grant number TWCF0268 (“Learning in evolution, evolution in learning”). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.