Selection by consequences, causality and essentialism: comments on Leão and Neto

Selección por consecuencias, causalidad y esencialismo: comentarios sobre Leão y Neto

The target paper (Leão & Neto, 2018; “L&N” henceforth), if I understood correctly, is a conceptual-historical analysis that downplays the possibility of an incoherence in Skinner’s early thought between his view of causation and an implicit, incipient form of his thesis of selection by consequences (SbC) he allegedly held at the time. According to SbC, operant conditioning is literally a process of evolution by selection at the ontogenetic level. L&N also present their analysis as an attempt to clarify the early development of Skinnerian psychology. The paper is thought-provoking and discussions of the topic can help clarify some issues.

The possibility of an incoherence, L&N assert, arises from the fact that Skinner (1931) defined the reflex as a necessary relation (p. 446), which allegedly carries a commitment to a mechanistic view of causality. This view, L&N and others claim, following Skinner (1981), is incompatible with the thesis of SbC. According to Skinner (1981), SbC “is a causal mode found only in living things” (p. 503) that “replaces explanations based on the causal modes of classical mechanics” (p. 501). At the time of his 1931, he also viewed the operant as a reflex (which, of course, he changed later on).. All this leaves his early writings open to the accusation of incoherence, which seems to the main concern of the target paper.

No need to worry, though, at least about such an accusation, or so L&N argue. They provide textual evidence of a transition in Skinner’s thought from a mechanistic to a selectionist view of causality. Scharff (1982) made a similar point in her analysis: Skinner shifted from “necessitarian” (mechanistic) to probabilistic causality, and abandoned the view of operants as reflexes (but see Note 2). These transitions suffice to refute, or at least downplay, any accusation of incoherence against Skinner’s early thought. By definition, a transition is an indeterminate, noncommittal intermediate state that does not lend itself to a definite accusation of incoherence between strongly-held, well-defined positions.

Still, there are good reasons to believe no incoherence existed to begin with in Skinner’s early thought between a full (even if momentary) commitment to mechanistic causality and any implicit, incipient form of the thesis of SbC he might have held at the time. Moreover, Skinner’s shift to probabilistic causation pointed out by Scharff (1982) was not required to ensure coherence with the idea of SbC. I am saying that the claim of an incompatibility between mechanistic causality, the Darwinian theory of phylogenetic evolution by natural selection and, to this extent, operant conditioning viewed as SbC, is mistaken. Let me explain.

As I said in the concluding remarks somewhere else (Burgos, in press), no defender of SbC has clarified exactly what is the causal mode of SbC or classical mechanics, or why the latter is inadequate for SbC. In what follows, I assume that by “classical” Skinner meant “Newtonian” (if he did not, I have no idea what he meant, in which case, I have nothing else to say about the matter; if he meant “Cartesian,” his rejection of the causal mode in “classical” mechanics was vastly outdated). I also assume that he accepted the conventional (but, as I argue below, mistaken) wisdom according to which Newtonian mechanics commits us to the causal mode of efficient causation. If both assumptions are correct, the case for SbC hinges on a rejection of Newtonian mechanics as inadequate for a science of behavior (and evolutionary biology), because the former commits us to efficient causation and this is an inadequate causal mode for a science of behavior (or evolutionary theory). As usual, however, the matter is much more complicated.

For a start, Newtonian mechanics is entirely compatible with probability theory. The combination of the two with classical thermodynamics gave rise to (nonquantum) statistical mechanics, a foundation of modern physics. Thus, there is no such opposition between mechanistic and probabilistic causality. Let us not confuse mechanistic with strictly deterministic causality where the cause always brings about the effect. The existence of probabilistic machines clearly illustrates the compatibility: Something can be a machine without being strictly deterministic. The statistical nature of evolution by natural selection and operant conditioning, then, is no excuse to repudiate explanations of them by mechanistic causality.

An example of a Newtonian-mechanistic explanation of natural selection with probability considerations is Sober’s (1984) interpretation of natural selection as a Newtonian force that acts upon evolution, viewed as a change in the gene frequencies of a population. Mutation, migration (both analogous to gravity), and genetic drift (analogous to Brownian motion) are others. Sober (1984) thus proposed that the theory of evolution consisted of three kinds of laws: A zero-force law (which describes how evolution occurs in the absence of any forces, natural selection included), source laws (which describes how the forces of evolution, e.g., natural selection, are produced), and consequence laws (which describe how forces, if present, drive evolution, i.e., the effects of the forces).

According to Sober (1984), the Hardy-Weinberg equilibrium is the zero-law of the theory of evolution, as it describes (final, steady-state) gene frequencies as a result only of allele frequencies, without the effect of any force. The law can be adjusted to take into account the presence of these forces, either in isolation, or in composition. Sober (1984) formulates all this purely in terms of Newtonian force vectors, where such forces push gene frequencies into particular directions, very much in the same way that forces in Newtonian mechanics push particles and deviates them from a straight line. The interpretation is coherent, very clear and precise, and allows for a more rigorous articulation of the theory of evolution.

Equally erroneous is the suggestion that operant conditioning is unexplainable by mechanistic causality. The theory of behavioral momentum (Nevin, Mandel, & Atak, 1983) uses Newtonian mechanics to explain resistance to change in operant conditioning in terms of momentum as the product of mass and velocity. Killeen (1992) has also explained operant conditioning in terms of Newtonian forces, within a behavioral selectionist framework! Such efforts show that operant conditioning qua SbC is explainable in terms of Newtonian mechanics. This outcome is at odds with Skinner’s (1981) recommendation that the causal mode of SbC replaces the causal mode of Newtonian mechanics, a recommendation that suggests an incompatibility between the two, but perhaps I am missing something, so I must delve deeper.

As I mentioned, conventional wisdom has it that the causal mode of Newtonian mechanics is efficient causation, the mode that comes to mind most naturally in relation to motion by direct collision. Efficient causation is incompatible with Skinner’s (1981) thesis of SbC in that efficient causation is a strictly deterministic and, hence, necessary relation, of the sort observed in unconditioned reflexes, but not in emitted responding in operant conditioning, the focus of SbC (see Note 2). However, this interpretation of the causal mode of Newtonian mechanics as efficient causation is moot. There is no evidence that Newton reduced all Aristotelian causes to efficient, and some have argued that he did not. According to Faulconer (1995), Newton, influenced by Francis Bacon (1561–1626), viewed the objects of science as formal causes:

The essence of the two causes Newton accepts, mechanical and first, is that they are both formal causes: each is the law of a science, a formal description of the origin of motions. ... In being mathematical, the mechanical is formal. Even though the word mechanical suggests to us strongly that Newton’s physics is a matter of materialistic determination, that suggestion is misleading. Newton’s physics is mechanical because it deals with force; the word mechanical is a technical term in Newtonian science: to be a mechanical cause is to be a force that is mathematically describable. ... Though it is common to think of Newtonian causation as efficient causation and to say that modern science reduced Aristotle’s four causes to two, material and efficient, it is more accurate to say that Newton reduces Aristotle’s four causes to one, formal cause (p. 81).

This interpretation could still support Skinner’s (1981) suggestion that the causal mode of classical mechanics is incompatible with the causal mode of SbC, but in a different way. This possibility leads to the topic of essentialism, over which there is much confusion in the behavioral selectionism literature. Unfortunately, L&N perpetuate the confusion.

The possibility in question arises from interpreting Aristotelian formal causes as essences (a standard albeit not the only interpretation), which makes Newtonian mechanics essentialistic. According to Palmer and Donahoe (1992), however, whom L&N cite to support their analysis, the thesis of SbC, as a form of selectionism, is anti-essentialistic. This position would seem to be at odds with a Newtonian treatments of evolution and conditioning. Fortunately, the opposition is only apparent and can be easily dispelled in favor of my initial conclusion that Newtonian mechanics is entirely compatible with SbC.

One key clarification is that the only essentialistic thesis rejected in population thinking in evolutionary theory is the Aristotelian (typological) concept of species, according to which they themselves are essences (e.g., Mayr, 2001). The problem with this concept is clear: Essences do not change, but species do; therefore, species cannot be essences. However, Sober’s (1984) Newtonian-mechanistic reconstruction of evolutionary theory in no way implies that concept of species. The former is entirely compatible with a rejection of the latter.

Still, some might retort that the essentialistic character of Newtonian mechanics prevents it from describing change, which is key (essential?) to evolution and operant conditioning. Such an objection, however, would be equally mistaken. Newtonian mechanics is all about change. Its method of fluxions and fluents (nowadays known as calculus) was designed to describe change mathematically. A dynamical function . is a mathematical description of how a certain magnitude . (the dependent variable) changes in time (., the independent variable). The derivative of a dynamical function (^dx_dt), defined as the slope of the tangent line at a point of the function, is a measure of the instantaneous rate of change at that point.

How can Newtonian mechanics describe change and be essentialistic, if essences are unchangeable? The answer is that what is unchangeable in Newtonian mechanics are the laws of motion, which constitute the essence of motion. The Second Law asserts that . = ma (. denotes “force,” . “constant mass,” and . “acceleration”). This law is unchangeable, even if it describes change (. is a measure of the rate of change of velocity, the second derivative of position as a function of time). This essentialistic character of Newtoninan mechanics is thus entirely compatible with change.

An equation that describes changes in gene frequencies from one generation to the next due to the effects of migration, mutation, and/or selection is unchangeable in the same sense: The equation itself is does not change, even if it describes change. The very same equation is used to calculate different rates of change, which implies that the same law is at work: The law itself (just like the three laws of motion in Newtonian mechanics) does not change. In the case of operant conditioning, a mathematical description of resistance to change in the behavioral momentum theory is unchangeable in the same sense: The description itself does not change; what changes is its various numerical solutions. And so on.

This form of essentialistic thinking, where immutable laws are viewed as essential to dynamical phenomena, is fully compatible with change and population thinking in evolutionary theory. There is no incoherence in speaking of the immutability of laws that determine rates of change as essential to dynamical phenomena or processes such as evolution and operant conditioning qua SbC. These forms of essentialistic thinking do not entail the essentialistic (typological, Aristotelian) concept of species.

There is yet another form of essentialistic thinking more directly related to the concept of species, but does not imply the typological concept of species either and, hence, is compatible with evolutionary theory: The notion that species have essences or essential properties. One thing is to say that species are essences (the typological view, untenable under evolutionary theory), quite another that species have essences. Having essential properties is entirely compatible with change in ways that satisfy the theory of evolution by natural selection. In particular, the formation of a new species, known as speciation, is the coming into existence of an individual with certain essential characteristics that remain fixed while the individual changes.

All of this allows for meaningful talk of the same species through time, underlying uses of biological taxonomy. Homo sapiens sapiens has been the same species since it formed about 315,000 years ago, even if it can be said to have evolved and still is evolving in a standard sense (e.g., changing of gene frequencies of its population). We do not say that it is a different species just because it has evolved. Only in speciation can we correctly say that a new species evolved, but this is not the only sense in which species can be said to evolve. What allows such talk to make sense is the assumption that our species has some essential properties that remain (roughly) fixed through time, and which define it. Exactly what they are, of course, remains a matter of debate, but many biologists consider bipedalism, opposable thumbs, an articulate language (often viewed as an exaptation), and a certain genotype as defining (most likely fuzzily) of our species.

All in all, then, the only form of essentialistic thinking that is incompatible with evolutionary theory is the typological concept of species. Other, more widespread forms (viewing laws as essential to phenomena, defining species in terms of essential properties) are compatible with evolutionary theory. I thus stand by my initial conclusion: Skinner (1981) was mistaken in suggesting that the causal mode of classical mechanics was incompatible with (to be replaced by) the causal mode of SbC.

But all this is much ado about nothing if intended to support SbC. The target paper, albeit not strictly a defense of SbC, has not changed my mind about SbC (see Note 1). Nor has a recent defense (Stahlman & Leising, in press). If I am right in my rejection, L&N’s analysis is about how a blunder came to be, an important contribution. Conceptual-historical analyses of the phlogiston, the luminiferous aether, geocentrism, and flatearthism can be very valuable, especially to identify where the faulty steps were taken, as morals to prevent us from repeating them.

[Artículo comentario , vol. 44, 1] http://dx.doi.org/10.5514/rmac.v44.i1.65349

Correspondence concerning this article should be addressed to Centro de Estudios e Investigaciones en Comportamiento, Universidad de Guadalajara. Av. Fco. De Quevedo 180, Arcos Vallarta, Guadalajara, Jalisco 44130, Mexico. jburgos@cucba.udg.mx, http://www.ceic.cucba.udg.mx/. I thank Kennon A. Lattal for his kind invitation to write this commentary.