2.1 Saltation, emergence and non-linear dynamics

Hugo De Vries rediscovered the Mendelian model of genetic inheritance in the late 19th century. Early 20th-century genetic research (Morgan 1916; De Vries 1917) suggested that heritable traits were discrete. There were dominance-relations that could silence the expression of genes and polymorphisms that could protect deleterious gene-combinations from extinction. Many biologists saw this as a death-blow for the Darwinian model. The experimental evidence suggested: 1) that small-scale variability was often not heritable and 2) that heritable variation was quantised, so saltatory dynamics were probable.

The modern synthesis provided an empirically defensible alternative to Darwin's model, but struggled to accommodate later developments in systems ecology, many of which revolved around the concept of emergence. For a discussion of the relationship between emergence and the extended synthesis see Pigliucci (2014). Again, we will use language anachronistically to tease out what, to us, appear to be the key ideas.

In 1923 C. Lloyd Morgan, one of Thomas Henry Huxley's students, wrote a book titled Emergent Evolution. Morgan borrowed the word 'emergence' from the 19th-century philosopher George Henry Lewes (1875, vol. II. Prob. V. ch. iii, 369). Lewes distinguished two broad types of logical relationship. The first, following John Stuart Mill (1843, Bk. III. ch. vi. § 2), Lewes called resultant; the second he called emergent. We illustrate the difference by example.

Imagine an experiment in which a scientist uses a catapult to launch a small glider. A strong impulse moves the glider a long way and exposes it to wind currents longer than a weak impulse, but it seems intuitively obvious that the long flight can be decomposed, as it were, into a sequence of short flights, which could be executed to much the same effect. There would be some statistical discrepancies between the outcomes, but the whole and the aggregate of the parts would be broadly equivalent. Resultant systems like this one are time-symmetric; their behaviour can be explained ex post and predicted ex ante because it can be decomposed into small steps and extrapolated, step-wise into the future.

Now imagine a scientist making many of these experiments, in one of which a dog chases the glider, picks it up and runs away. The scientist tries to retrieve the plane and the dog takes evasive action. Two things have happened: first, the outcome of this particular experiment clearly belongs to a different ontological class. The scientist/glider system has self-organised into a completely new type of dynamic. Second, that symmetry-breaking event means that the whole is no longer decomposable into the sum of its parts. The common-sense assumption of resultance (we would now say 'linearity') must be set aside. Lewes called these non-linear systems emergent. C. Lloyd Morgan used this distinction to explore a number of contentious ideas in evolutionary theory, including the question of agency among animals and the relationship between science and theology.

Emergent phenomena have two important properties: first they are time-asymmetric - past system behaviour is a poor guide to future dynamics; second, they represent a shift from one attractor to another. These two manifestations of emergence often correspond to different space-time perspectives. The winning numbers in a lottery, for example, are unpredictably emergent in the sense that they cannot be predicted ex ante. Viewed ex post, however, the effects of the symmetry-breaking event on those who possessed the winning ticket can be explained in terms of a self-organising transition.