Genes, Phenes and the Baldwin Effect: Learning and Evolution in a Simulated Population

Abstract

Genes, Phenes and the Baldwin Effect: Learning and Evolution in a Simulated PopulationbyRobert M. French and Adam MessingerWillamette University, Salem, OR 97301(french@willamette.edu and amessing@willamette.edu)IntroductionThe view that evolution is influenced by acquired behaviors and traits is regardedby many as being uncomfortably close to the discredited Lamarckian contention thatevolution consists of the inheritance of acquired behaviors and traits. It is perhaps for thisreason more than any other that the evolutionary mechanism first proposed first by J.Mark Baldwin and Lloyd Morgan in 1896 [Baldwin, 1896; Morgan, 1896] is still veiledin controversy. This mechanism, known today simply as the Baldwin Effect, states thatlearned behavior and characteristics at the level of individuals can significantly affectevolution at the level of the species. Schull [1990] sums up the process as one in whichindividual developmental responses will necessarily lead to directed and non-randomevolutionary change. And while many evolutionary biologists accept the Baldwin Effectas a significant force in evolutionary change, the theory also has many detractors. Forexample, in a recent article, Piattelli-Palmarini [1990] writes, One would have hopedthat, in 1990, all talk of the Baldwin effect . . . would have been mercifully forgotten.Parisi, Nolfi, and Cecconi [1990] give three further reasons that evolutionarybiologists tend to dismiss the Baldwin Effect. The orthodoxy of evolutionary biologists,they claim, is strongly reductionist, “which implies that the causes and basic mechanismsof evolution are only to be found at the level of genetics.” As a consequence, behaviorand learning, both being highly holistic processes, have been largely ignored inattempting to understand evolutionary processes. Another reason for the lack of attention,according to these authors, is that evolutionary biologists feel “behavior and learning arethe province not of biology but of psychology and ethology.” And finally, until recently,there have been very few empirical studies of the Baldwin Effect in either real orsimulated populations. In this paper, clear evidence is presented that the Baldwin Effect can indeedsignificantly alter the course of Darwinian evolution at the level of the genotype. In otherwords, we show that plasticity at the phenotypic level can and does produce directedchanges at the genotypic level. In addition, the amount of plasticity and the amount ofbenefit of the learned behavior are also demonstrated to be crucial to the size of theeffect: either too little or too much and the effect disappears or is significantly reduced.Finally, we demonstrate certain conditions under which the Baldwin Effect is morepowerful in sexually reproducing populations than in asexually reproducing ones. Thisresearch confirms and extends earlier experimental work done by Hinton & Nowlan[1987], Belew [1990], Parisi, Nolfi, & Cecconi [1992], and Fontanari & Meir [1990],among others.