Emergence is not ordinary change: Introduction to Baylis

The idea of emergence in its complexity science sense was first broached by adherents of Emergent Evolutionism, a loosely joined movement of scientists, philosophers, historians, social thinkers, and even theologians during the first quarter of the twentieth century (Blitz, 1992). Included among the proponents of Emergent Evolutionism in England were the animal behaviorist C. Lloyd Morgan and the philosophers Samuel Alexander, C. D. Broad, and Alfred North Whitehead (the eminent mathematician had turned to philosophy in his later years). In the US, Emergent Evolutionism was promulgated by the entomologist W. H. Wheeler (who incidentally was one of E. O. Wilson’s teachers), the philosophers Roy Wood Sellars and John Boodin, the philosopher and historian Arthur Lovejoy, and the social theorist and philosopher George Herbert Mead (quite influential to the esteemed contemporary philosopher Jürgen Habermas).

Baylis’s four types of emergent change

Not surprisingly, criticisms of the concept of emergence began right on the heel of this movement. A common objection was voiced against the claim of emergence being impervious to reductionist or mechanistic explanations. One strategy of the detractors along this line was to take on those purported characteristics of emergent phenomena that enabled it to be protected from reductionist onslaught, e.g., the property of radical novelty of emergent level phenomena in relation to the lower, substrate level (Goldstein, 1999). Such was the tack taken in 1929 by the American philosopher Charles Baylis in the following article. In particular, Baylis, who taught at Brown and the University of Maryland, believed that he had, by puncturing the claim of radical novelty, demonstrated that emergence was no different than the type of discontinuity accompanying ordinary change and therefore could be explained by causal mechanisms that were typically applied to ordinary change.

To advance his argument, Baylis argued that there were only four possibilities of emergent change:

  1. Integrative emergence: when a new property appears as the result of an integration of components, e.g., at room temperature, hydrogen and oxygen are gases yet when chemically combined the result is the new property of liquidity;

  2. Integrative submergence: when properties of the parts are lost in the resulting whole, e.g., water (again) in which the gaseous property of the component elements is discontinued in the liquidity of the whole;

  3. Disintegrative emergence: when new properties of the parts emerge when they are separated from the whole of which they were previously a part, e.g., what happens when oxygen and hydrogen are separated from water;

  4. Disintegrative subemergence when the properties the parts had when combined are lost when they become separated from the whole, e.g, when hydrogen and oxygen are separated from water and thus lose property of liquidity. (To be sure, points 3 and 4 appear to be two sides of the same coin).

For example, Baylis referred to the complex (a+b), which, having the character of being greater than both a and b alone, could be considered an instance of ‘integrative’ emergence since it had a character neither of its elements had alone. Furthermore, if this same complex were dissolved, its unique character as a complex would ‘disintegrate’ thus losing the previous properties it had as a whole complex. However, and this was a key move on the part of Baylis, from another perspective, the supposedly integrative complex could at the same time be seen as disintegrative in that the properties that a or b alone might have had before they were combined into the complex (a+b) would be lost when a and b became part of the complex (a+b). Effectively, what Baylis had considered was how an emergent whole could be considered less than the sum of its parts!

Since these four types do indeed appear to have encompassed all the possibilities of what can take place during changes from parts to wholes and vice versa, Baylis was arguing that the discontinuous novelty characterizing emergent phenomena was ubiquitous to all changes and therefore it was arbitrary to identify emergence only with the first type, integrative emergence. To drive his point home, Baylis considered the change resulting from moving a book from one shelf to another. The change of place of the book could be considered either an integrative or a disintegrative novelty depending on your point of view: a new integration exhibited in the novel pattern of books at the new location or a new disintegration of the pattern on the shelf before the book was moved. Baylis concluded that novelty was thereby constituted by either a building-up of integrations or their tearing down. Consequently, the production of discontinuous novelty by itself didn’t aid our understanding of so-called emergent evolutionary processes.

Furthermore, since it was arbitrary as to whether a given change was to be considered an integration or a disintegration, Baylis believed the erection of any particular hierarchy of putatively emergent levels was an entirely discretionary matter, emergentists merely picking out and hierarchically arranging just those levels which interested them according to their own agendas. Because novelty was so pervasive with all change, moreover, a new level could be assigned every time something new came about and thus leading to a pervasive infinity of levels, an idea which Baylis found absurd. Or as the philosopher C. W. Berenda (1953) once blithely put it in commenting on the idea of emergence in the nineteen fifties, “There is nothing new in novelty”!

Stace’s causal chain

We can take Baylis’s argument, however, in another way, namely as having set the bar higher as to what should count as genuinely emergent novelty and not merely that kind of novelty which accompanies ordinary change. That is, by pointing out the arbitrariness as to what to count as emergent novelty, we can take Baylis as having demonstrated that the early emergentists had not sufficiently qualified emergent novelty. This suggestion will become clearer by looking at a similar argument offered around the same time as Baylis by another critic of proto-emergentism, the philosopher W. T. Stace (1939; incidentally, Stace achieved his fifteen minutes of fame much later for arguing against the religious authenticity of psychedelic drug-induced mystical experience, see Stace, 1960). Similar to Baylis’s concern with discontinuity during change, Stace’s dispute with emergence focused on the nature of causal chains operative during any process of change. In the causal sequence A…B, if B was discontinuous with A, then there was something new by definition in B. But, since every time there was a causal sequence, the focus could be either on how the effect was novel or how the effect was similar with respect to the cause (since causality implied both a difference and a similarity between cause and effect), Stace held emergentists to the task of justifying why they pushed for emergent phenomena being novel and not the same as antecedent conditions. He concluded that the only reason for emphasizing discontinuity had to do with the relative commonness or rarity of an experience: “…when [emergents] do occur they seem more striking and unexpected, and so we get the impression that we are in the presence of some kind of novelty which is absent in the other cases” (Stace, 1939: 308). Indeed, many unexpected things can happen, but their unexpectedness might merely be old things happening in unexpected places. The sense of unexpectedness was, after all, a subjective attitude on the part of an observer’s mind, a sense of surprise projected into a claim of novelty. The sense of novelty, for Stace, therefore said more about the person beholding an emergent than the emergent itself. Moreover, according to Stace, novelty as such was compatible with many philosophical positions, deterministic and indeterministic, mechanism and vitalism: “For all I know the world may be alive, organic, creative, soulful, poetical, noble. But the concept of novelty does not help us to show that it is so” (Stace, 1939: 310). As a result, Stace held that novelty was merely epistemological and had no real bearing in establishing the ontological status of emergence.

The contextuality of emergent novelty

The arguments of both Baylis and Stace can be explicated along the lines of the mathematical construct of equivalence classes in which an operation is devised and a definition of equivalence and inequivalence is posited. The equivalence class is constituted by those members defined to be equivalent for that particular example, but it is an arbitrary designation depending on how one defines the operation and the equivalency. For example, in the operation known as congruency, one number is defined as equivalent to another according to whether it has the same remainder after being divided by some arbitrary number (Ash, 1998): if the operation is defined as division of an integer by 4, then one possible equivalence class would be those multiples of 4 that when divided by 4 leave a remainder of 0, e.g., the class made-up of (…-8, -4, 0, 4, 8 …). Notice that equivalency in this context does not imply identity which would be the complete absence of novelty. That is, -8 is not identical to 4 but, instead, for the particular operation of dividing by 4 and resulting in 0 as a remainder, -8 and 4 are equivalent, or, what amounts to the same thing, not different with respect to that operation. The point is that what is considered the same, i.e., equivalent, and what is considered not equivalent, i.e., novel, are arbitrary markers depending on how one has first defined what is to count as equivalent or not.

Besides the element of arbitrariness involved in deciding what’s novel, there is another aspect of the contextual nature of novelty, namely, that something is either novel or similar with respect to another thing according to how that same thing is novel or similar with respect to another thing altogether. For example, British English and American English are distinct, i.e., novel with respect to each other on many points, yet they are more similar to each other than they both are to French. Similarly, English is more similar to French than French or English are with respect to Japanese. The point again is that the designation of something as novel is contextual and this must be made explicit when specifically considering emergent novelty.

Getting back to the point made above about taking Baylis (and now Stace) as having set the bar high for what is to count as genuinely emergent novelty, it must be recognized that a serious question-begging flaw can be detected in arguments that go from the contextuality involved in defining novelty to the dismissal of emergence by being identical with ordinary change. As a matter of fact, both Baylis and Stace started off their discussions of emergence by first identifying it with ordinary change in general: Baylis did this by conflating emergence with his four types of change and Stace by understanding emergence within the framework of change as a causal chain. But once these identifications have been made, then, of course, any novelty characterizing emergent phenomena will essentially be the same as the novelty accompanying change in general.

In any effect, we now have a bar over which emergence must be defined: emergence is not ordinary change in general but is instead consonant with a special kind of change, i.e., one that generates the outcomes which are unpredictable, non-deducible, irreducible, and capable of daunting (not violating) traditional notions of causality and determinism (Kekes, 1966). That is, if the construct of emergence is to indeed have the significance emergentists want it to have, it must be referring to a very different kind of change than the ordinary kind. As the philosopher Paul Henle (1942) pointed out over fifty years ago, the novelty found in doctrines of emergence amounted to much more than, say, a new automobile coming out of an assembly line in that not just the actual matter making up a car itself must be new, the form must also be novel.

A failure of the imagination

The criticisms of emergence leveled by Baylis and Stace both can be said to have rested on a similar picture of the process of natural change. We can see Baylis’s picture in what he appealed to in order to explain how novelty could be associated with every change: the process of moving a book from one shelf to another. To be sure, such a change does fit with Baylis’s definition of the novelty of emergent ‘complexes’ which he equated with the ‘gestalts’ of Gestalt psychology. The process of moving a book from one shelf to another leads to not only a new pattern on the new shelf, but a new pattern on the old shelf. But there is nothing unpredictable or irreducible about these two instances of novelty. The book is easily taken back from its new place to the old one, thereby, obliterating the novelty the original movement brought about. By understanding emergence in the specific terms of how he conceived change and novelty, Baylis could maintain that there was both a type of symmetry and arbitrariness between integration and disintegration. Moreover, Baylis’s process of emergence thereby shared the attribute of reversibility. Similarly, Stace’s simple picture of the causal chain had no place in it for the nonlinear, complexifying operations going on in complex systems. In both cases we see a failure of the imagination as to what is possible for nature. In the age of complexity, such a failure of the imagination can no longer get by without notice.


Originally published as Baylis, C. A. (1929). “The philosophic functions of emergence,” The Philosophical Review, ISSN 0031-8108, 38(4): 372-384. The E:CO editorial team would like to thank Duke University Press, especially Thomas Robinson, for their kind permission to reprint this article. The original article can be downloaded from here.