Essay:What Kind of “Cause” is Natural Selection?

From RationalWiki
Jump to navigation Jump to search
Essay.svg This essay is an original work by OnlySortaDumb.
It does not necessarily reflect the views expressed in RationalWiki's Mission Statement, but we welcome discussion of a broad range of ideas.
Unless otherwise stated, this is original content, released under CC-BY-SA 3.0 or any later version. See RationalWiki:Copyrights.
Feel free to make comments on the talk page, which will probably be far more interesting, and might reflect a broader range of RationalWiki editors' thoughts.

Introduction[edit]

In the work Making Sense of Evolution: The Conceptual Foundations of Evolutionary Biology, the authors Pigliucci & Kaplan (2006) emphasize the limitations of various statistical methods used in evolutionary biology and population genetics as lacking the ability to determine the true causal mechanisms behind natural selection. Pigliucci & Kaplan (2006) use the metaphor of an Indonesian shadow puppet show to argue that using statistical methods to infer causality when it comes to natural selection is like trying to infer the structure and mechanisms of the puppets from their shadows.

Of course, in this mention of the causal mechanisms of natural selection, causality itself is left unanalyzed. The goal of this essay is to provide an account of causation in evolutionary processes, so as to identify what it is exactly that needs to be sought in studying the causal mechanisms of natural selection.

Pigliucci & Kaplan’s Critiques of Popular Statistical Methods[edit]

Grow-Ratio Comparisons[edit]

When comparing a trait’s impact on fitness, evolutionary biologists will compare the growth-ratios of relevant populations based on the presence or absence of a given biological trait (Pigliucci & Kaplan, 2006). It is thought that differences in growth-ratios between populations of the same species within the same environment can be explained by differences in fitness between the populations based on the trait’s presence or absence. A known limitation of this method known to evolutionary biologists as mentioned by Pigliucci & Kaplan (2006) is that differences in growth-ratios can just as readily be explained by genetic drift as they can by natural selection. There is no way to determine without experimental intervention whether a trait is being selected for, or is merely the consequence of genetic drift (Pigliucci & Kaplan, 2006). Pigliucci & Kaplan (2006) also come to emphasize that genetic drift itself is not a physical or causal process.

Lande & Arnold’s Multiple Regression Analysis[edit]

Lande & Arnold’s method of multivariate analysis appreciates widespread use in quantitative genetics according to Pigliucci & Kaplan (2006). The analysis essentially looks at a set of covariable phenotypic traits and determines what is called a directional selection gradient (β) to assess which traits have been selected for given known selection pressures from the environment (Pigliucci & Kaplan, 2006). The original developers Lande & Arnold interpreted β as a set of coefficients that approximate “the fitness surface”, and when plugged into a multivariate breeder’s equation, can be used to make quantified predictions about how a population responds to selection (Pigliucci & Kaplan, 2006, pp.48). Pigliucci & Kaplan (2006) allege that Lande & Arnold make a series of problematic assumptions to reach this conclusion, namely in assuming that the traits will be statistically distributed in such a way as to be said to be “multivariate normal”, and that the traits themselves will not exhibit such a strong correlation as to produce “unstable” outputs (pp.49). Something acknowledged by Lande & Arnold themselves is that the method of using multiple traits and accounting for their fitness as independent variables is only as reliable as the set of covariant traits being measured is complete – which is highly difficult to determine (Pigliucci & Kaplan, 2006). It should also be noted that this particular type of analysis incorporates the use of heritability (or at least the multivariate conception known as G) which comes with its own limitations.

Heritability (h²)[edit]

Despite common assumptions among the lay public, heritability (h²) does not actually measure the genetic contribution to a physical or behavioral trait (Moore & Shenk, 2016). All heritability (h²) measures is the degree to which phenotypical variation is related to by genotypic variation in a given population in a specific environment (Moore & Shenk, 2016). It can not tell you how a trait is affected by genetic factors, if different populations differ in phenotypic traits due to genetic differences, or how the trait itself is inherited. All heritability amounts to mathematically is simply….

h² = Vg/Vp

Where Vg is the degree of variation in genotype for a population, and Vp is the degree of variation in phenotype for a population (Pigliucci & Kaplan, 2006).

Pigliucci & Kaplan (2006) frame heritability (h²) as a measure of covariance between phenotype and genotype, and put extra special emphasis to which the degree of heritability (h²) changes dramatically depending on the environment that particular population is being measured within. It has utility in the context of animal and crop breeding but that is due in part because of the degree to which the organism's environment is carefully and artificially controlled (Pigliucci & Kaplan, 2006). For long term predictions regarding a population’s response to selection pressures, it is not so useful.

G-matrices[edit]

Heritability (h²) is utilized in what is called the breeder's equation, and Lande & Arnold proposed a multivariate version of the breeder's equation that instead utilizes the concept of G (Pigliucci & Kaplan, 2006). G denotes a kind of matrix that lists the covariances and heritability (h²) between different traits along two axes, indicating the degree to which an individual trait is heritable and the degree it is statistically correlated with other traits within the matrix (Pigliucci & Kaplan, 2006). G is the multivariate equivalent to heritability (h²) and so all the relevant caveats and limitations of heritability (h²) carry over to G (Pigliucci & Kaplan, 2006).

Various problems listed by Pigliucci & Kaplan (2006) that they identify in the relevant literature is that like heritability (h²) G is subject to change depending on changes in the environment and the population looked at; the measure itself is not stable enough to work over long periods of evolutionary time; it is not known how evolution impacts or changes G; and because different genetic architecture can result in the same G it can not be used to indicate causal pathways between the development of a given trait and the relevant genetic resources.

With all that, Pigliucci & Kaplan (2006) strongly emphasize that the use of G in quantitative genetics be used in limited contexts, and for researchers not to draw strong conclusions from such measures.

The general theme[edit]

In their criticisms of various statistical methods used in population genetics and evolutionary biology, Pigliucci & Kaplan (2006) seem to suggest that there is something physically interacting between the environment, populations, individual organisms, their physiological traits, and genotype that all result in the phenomena we call natural selection. The authors seem to want to emphasize that despite this – statistical and quantitative analysis is ill-equipped to explain how these interactions work causally.

Theories of Causation and Natural Selection[edit]

If one wishes to identify the causal mechanisms underlying natural selection and evolution it stands to reason that we need an adequate description of what causal mechanisms themselves even are, or put another way – we need a metaphysical theory of causation.

Skepticism of Causality, and Causation as Constant Conjunction[edit]

No philosophical discussion regarding causation would be complete without mention of the Scottish philosopher David Hume’s (1711-1776) skeptical argument against the existence of causation.

In his work, An Enquiry of Human Understanding published in 1748, Hume famously casted doubt on the concept of causation being a matter of “necessary connection”; i.e. as something where A causes B on the basis that some force or property in event A necessitates the result of event B (Hume, 2008). Hume understood this to be how we intuitively think of causation and how the rationalists of his time characterized it, but Hume argued we have no empirical basis or method of demonstrable proof that this occurs. This “force”, “property”, or “necessary connection” was not something we can perceive, and it was not something we can prove existed non-circularly with the use of reason; and so, on that basis, Hume concluded we had no rational basis to assume any such causality existed (Hume, 2008,).

Hume in essence argued that causation itself could not be derived from sense experience, nor could it be proven to occur, as it could just as easily have been that two uncaused events happened in close proximation to each other; we could not prove that this was not the case (Hume, 2008).

Hume (2008) proposed that what we mistook for necessary connection, or “causation”, was merely the constant conjunction of A and B happening together, or the near-constant occurrence of B directly after A. Thought relevant to Hume was that events had to occur in spatiotemporal proximity to each other. Some of the Logical Positivists took this as a direct theory of causation, that a causal event was merely the constant conjunction of two or more events in spatiotemporal proximity (Psillos, 2002).

As a theory of causation, it would have to be rejected, as it does not distinguish between mere correlation and cause. It had been said that the philosopher Immanuel Kant (1724-1804) was so punctual and consistent regarding his daily walks that his neighbours would set their clocks according to when Kant would start his walk. We can use that alleged fact to motivate a counter-example.

Say for the sake of argument we have a woman named Sarah who has recently bought herself a clock. Upon putting up the clock in her front foyer, and turning it on, she decides that she will go for a walk at exactly six o'clock in the evening, and in the morning, every single day. As an unconscious habit every time she is preparing to go for her walk she places her hand on the outer rim of the clock. Every day at the exact moment at six she places her hand on this clock while preparing to go for her walk. For unrelated reasons on days she is unable to go for a walk the clock would break before striking six. When Sarah passed away this clock would also mysteriously break, never striking six again. This would amount to a series of implausible coincidences, it could not be said that the clock striking six was causing Sarah to go on her walk, or that Sarah in preparing to go for her walk caused the clock to strike six. Yet if we are committed to the theory of constant conjunction we would have to conclude that yes, these events are causing one another due to their constant conjunction and spatiotemporal proximity to one another.

If Natural Selection were to possess any kind of causal mechanisms at all, this causality could not be characterized as a constant conjunction.

Hume’s skepticism relies on two principles, the absence of proof from demonstrable reasoning, and the absence of being able to empirically observe causation. If we were to demonstrate the reality of causation against Hume’s skepticism then we would need a description of causation that allowed it to be observed and readily identified. In that case, if causation was detectable empirically then it could in a sense be perceived. The absence of a proof from demonstrable reasoning isn’t presented as necessary or sufficient in establishing the rational justification for the belief in causation.

Lewis’s Counterfactual Theory of Causation[edit]

David Lewis (1973) argues that Hume proposed another conception of causation namely that A causes B if A were not to occur then event B would not exist. This becomes the inspiration for Lewis’s counterfactual theory of causation. The reference to counterfactuals requires explication as it carries with it certain metaphysical and logical implications. A counterfactual claim is one of the sort like “If Kennedy was not assassinated, then he would have had a wonderful end to his term” – if such a claim was interpreted as a material conditional the claim itself would be vacuously true simply based on the antecedent being false. For this reason, counterfactuals or subjunctive conditionals require a different sort of analysis.

Lewis (1973) provides an analysis of counterfactuals using modal semantics. A counterfactual claim like the sort “If A then B” has to be interpreted in terms of possible worlds. The worlds under consideration are only those comparatively similar to the actual world and the logical relation from A to B is of the form A ☐→ B which is defined by Lewis (1973) as true iff there simply are no possible worlds where A is true, or that in the nearest possible A-world B holds as true compared to any other A-world where B does not hold.

As a basis for causality, Lewis defines causation as the following…

“If c and e are two actual events such that e would not have occurred without c, then c is a cause of e.” (Lewis, 1973, pp. 563). 

With the counterfactual analysis provided above we interpret this “would not occur” as being the joint logical propositions of (c ☐→ e) and (~ c ☐→ ~ e.) (Lewis, 1973 ).

We can see the application of such reasoning in the way we may speak of natural selection. Take the example of insecticide-resistant insects. We may reason that the presence of insecticide, coupled with the inheritance of traits that resulted in organisms that do not react in the presence of this insecticide caused the increased presence of insecticide-resistance within the population upon further generations of insects. Without the use of insecticide and without the mutation that results in the particular trait of resistance being present in the population, we would not see an increased presence of insecticide resistance in the population. This would be an example of natural selection and counterfactual reasoning regarding causality.

This can be investigated scientifically when we compare populations of the same species exposed to a given selection pressure, and those who have not, while also trying to identify the related genetic and biomolecular factors that can give rise to the trait and identify its presence in the selection-exposed population compared to the non-selection-exposed population.

There is a problematic component in this suggestion as it is not at all obvious that this is a good approximation of varying counterfactual conditions without the utilization of appealing to other possible worlds as Lewis’s theory explicitly does. It could be argued that Lewis's theory provides the foundation to which true causality occurs, but that it can still be reasonably inferred from conditions of the actual world that closely approximate what is presumed in the counterfactual possible world as evidence of confirmation towards a causal mechanism.

Of course, Lewis’s counterfactual theory of causation is not without relevant objections, and if it turns out not to be an adequate theory for causation then it can not define what causal mechanisms are relevant to natural selection.

One particular issue for Lewis’ Counterfactual theory as originally conceived is the problem of preemption. Causal preemption is when two events are sufficient causes for a particular outcome, but they don’t happen at the same time. A thought experiment to illustrate the problem of early preemption is provided in the tale of Suzy and Billy throwing rocks; Suzy throws a rock at a bottle and it shatters; but if Suzy hadn’t thrown the rock then Billy would had thrown it instead and shattered the bottle with his rock (Collins, 2007). In this case, we may have a condition where the first associated proposition namely in (Throws Rock(Sally) ☐→ Shatters (bottle)), but not for the second proposition (~Throws Rock(Sally) ☐→ ~ Shatters (bottle)), as the bottle would still be shattered even in the nearest possible worlds that Sally did not throw the rock.

The philosopher John Collins (2007) believes this to be an easy fix for Lewis in changing the counterfactual theory so that it is confined to a particular chain of events that lead to the outcome of the bottle being shattered by Sally’s rock; so that, any other counterfactual consideration be restricted to that particular chain of events that caused Sally’s throw.

Setting aside the issue of preemption there is another thought experiment we can appeal to undermine the counterfactual theory of causation. Imagine if you will that we have two gunmen aiming towards a helpless victim. We can call one of the gunmen A and the other B. Our victim will be denoted as C. Both A and B fire their guns at the same time. The bullets from each gun fire through C’s heart at exactly the same time. In this case, we may be tempted to say that both bullets from the guns caused C’s death. If however counterfactually only one of the guns went off then C would still die and their death would not counterfactually depend on both gunmen hitting their target. If only A’s gun went off there is still the counterfactual case that B’s gun went off and killed C in the event that A’s gun had misfired. The inverse can also be true. In this case, we could not say that either A’s or B’s gun caused the death of C, nor could we say that A's and B’s guns caused the death of C.

Of course, the easy fix as referenced earlier by John Collins (2007) can be applied in this context but we can imagine a possible world where the causal chain of events stayed the same, but due to slight variations in the events after the guns had already fired one of the bullets missed their target. This brings us back to the same problem.

The tale of two gunmen and their one helpless victim gives us a case of causality that would not meet Lewis’s counterfactual theory of causation. This would mean that Lewis’s theory is insufficient as a theory of causation.

Salmon’s Process Theory[edit]

Wesley Salmon offered a direct theory of causation as an alternative to counterfactual theories. Salmon (1994) makes a distinction between what he calls causal processes and what he calls pseudo-processes, and goes further to make a distinction between causal processes and causal interactions. Casual interactions are said to be more “fundamental” than causal processes but the process typically would involve some kind of causal interaction (Salmon, 1994). Processes generally have the quality of being a continuity and place of intersection along a spacetime path but do not by themselves necessarily entail that an interaction has taken place (Salmon, 1994). Causal processes are distinguished from pseudo-processes in the presence of the capacity of a transmitted “mark” at the point of intersection; some form of altering information that affects the processes involved (Salmon, 1994).

Salmon (1994) puts this in terms of there being an interval along a spacetime path where two processes are altered by an intersection whereas the change persists beyond that point of interaction. This alteration is what Salmon (1994) calls a “mark”.

This notion of a “mark” and the distinction between causal and pseudo processes were not themselves coined by Salmon but originally by Reichenbach (Galavotti, 2018). Writing for the Stanford Encyclopedia of Philosophy Maria Galavotti (2018) explains Salmon’s view of causality in terms of the ability to transmit marks as what determines something to be a causal process, not that any actual transmission of a mark occurs.

The application of this to natural selection is not immediately obvious, nor if natural selection would even constitute a causal process in Salmon’s view. Salmon’s view of causality seems to suggest that there needs to be a specific spatiotemporal point to which the potential for transmission between processes can occur. Sometimes a “selection event” can occur such as a heavy rain storm, or a natural disaster that kills off large swaths of a population leaving only those with traits that were resilient to the event. These sorts of events are used as an example in the earlier mentioned multivariate analysis as an opportunity to measure a trait’s responsiveness to selection according to Lande and Arnold (Pigliucci & Kaplan, 2006). Introducing penicillin into a petri dish of bacteria arguably can be said to be a kind of selection event, but the effects of such an event typically take generations to complete.

It is not at all obvious that the introduction of penicillin constitutes a transfer of a “mark” into the population in precisely the way Salmon intended mark transference to mean, but there is arguably the transference of information into the environment and also upon the organisms themselves that results in their deaths of those who lack the traits relevant to antibiotic resistance.

It seems that causation as described by Salmon takes the form of objects on a given spatiotemporal trajectory intersecting one another and altering the characteristics of one another through a single direct interaction to be a causal interaction. Causal processes seem only to be when the potential of such an event takes place.

A favourite example from Salmon (1994) and reiterated by Galavotti (2018) is Salmon’s example of the spinning light in the center of a room. The light itself causes the appearance of the lit spot on the walls of the room and thus is a causal process, but the light spot itself spinning around the room is an example of a pseudo-process (Galavotti, 2018) (Salmon, 1994). Genetic drift could be characterized by this example as being an example of a pseudo-process.

Phil Dowe (2007) is of the stance that Salmon’s theory does not genuinely capture causal processes as what is included in Salmon’s account is that processes remain uniform when an interaction does not take place, in addition to, marks being transmitted without the presence of additional interventions from other objects. Dowe (2007) argues that even in the most ideal of circumstances causal processes happen with at least some presence of additional interventions, and so the vast majority of real-world causal processes would be ruled out by Salmon’s theory. This would especially be the case in the context of natural selection.

This would lead to the conclusion that Salmon’s process theory cannot be a true account of causality, let alone an account of causation as it occurs in evolution.

Dowe’s Conserved Quantity Theory[edit]

Phil Dowe (2007) presents his own theory of causation which he emphasizes as an empirical rather than conceptual analysis. In this context “causation” is analyzed in what empirical evidence suggests about cause and effect to provide an empirically accurate theory of causation with no regard to conforming to our intuitions about this topic.

Dowe (2007) presents two propositions for his conserved quantity theory of causation; those propositions being…

“CQ1. A causal process is a world line of an object that possesses a conserved quantity. CQ2. A causal interaction is an intersection of world lines that involves the exchange of a conserved quantity” (p. 90). 

The use of “world line” is a “collection of points on a spacetime (Minkowski) diagram that represents the history of the object” (Dowe, 2007, 90). A conserved quantity is essentially any type of quantifiable property that respects conservation laws described in natural science (Dowe, 2007). An example would be like a billiard ball travelling with a specific momentum and kinetic energy along a uniform path, until upon coming into contact with another billiard ball where part of that energy and forward momentum is exchanged into this other billiard ball. This alters the forward momentum and kinetic energy of the second billiard ball, while also changing the direction and lowering these values for the first ball. All the while the total amount of energy and momentum within this event is conserved.

An example from Dowe (2007) is in the event of a transmutation reaction such as when a nitrogen atom is struck by an alpha particle producing an oxygen atom and a hydrogen cation. In this case, the total electrical charge possessed by the atoms within the reaction is conserved. Another example Dowe (2007) uses is in radioactive decay where there is a change from one atom into two different types of atoms without being interacted with by another object, such as when radium-226 decays into radon and helium. The total electrical charge pre and post-decay is conserved, and the point where the atom splits into two characterizes an intersection given the split into two separate processes.

The application of the conserved quantity account (now referred to as CQ theory) to disciplines like chemistry and physics is readily identifiable as all chemical reactions and physical processes respect the conservation laws of energy, momentum, mass, charge, etc. There is not a single example within physics and chemistry where one variable causing another did not involve the exchange of energy, charge, or momentum. Either between interacting objects or at the very least the surrounding environment.

An objection to CQ theory comes in its seeming inapplicability to account for certain social and biological causes, i.e. the socioeconomic factors that govern wealth inequality, or the causal effects that smoking has on one's risk for lung cancer. In the context of evolution, it is not at all obvious what quantity or quantities are being conserved in the process of natural selection on a living population.

Dowe (2007) offers an additional theory for how we use the word “cause” in everyday language to talk of what is prevention and omission. Dowe (2007) does not believe prevention and omission to be genuine causes but believes they fit our use of “cause” in everyday language such as with statements like “drunk driving causes automobile accidents” or “alcohol causes cancer”. In the former example, alcohol can prevent normal functioning in brain cells which in turn omits reaction time and coordination – which then leads to a crash. Carcinogens can be said to prevent normal functioning in cells. Turning the lights off prevents visible light from filling the room.

In the context of evolution, we can see how this may play out in natural selection. A rabbit possessing a slow running speed due to natural variation is prevented from escaping predators, which in turn by resulting in death prevents the rabbit from producing fertile offspring. The absence of competition or obstacles to reproduction then omits any reason for faster rabbits to successfully reproduce.

Dowe (2007) provides the following analysis of prevention characterized as A causing Not-B ...

“A prevented B = A caused* Not-B if (P1) A occurred and B did not, and there occurred an x such that…(P2)  there is a causal relation between A and the process due to x, such that either (i) A is a causal interaction with the causal process x, or (ii) A causes y, a causal interaction with the causal process x, and (P3) if A had not occurred, x would have caused B” (pp,132).

Dowe (2007) emphasizes that even though he sees being caused* as not being an example of genuine causation the use of causation in P2 and P3 are referring to genuine causes as described by CQ theory.

This above analysis works for a selection pressure preventing the further reproduction of certain organisms in a living population.

Dowe (2007) provides a separate analysis for omission like so…

“Not-A caused* B if (O1) B occurred and A did not, and there occurred an x such that(O2) x caused B, and (O3) if A had occurred then B would not have occurred, and there would have been a causal relation between A and the process due to x, such that either (i) A is a causal interaction involving the causal process x, or (ii) A causes y, a causal interaction involving the causal process x” (pp.136).

Dowe (2007) thinks of omission as equivalent to had A occurred it would have prevented B. We can apply this in the context of natural selection when the absence of selection pressure allows for greater fitness.

But what would constitute x and y in the context of natural selection for the above analyses? Any example of the various biochemical and physical interactions within the organism and the environment that involves the exchange of conserved quantity could constitute the contribution of the variables x and y, which is difficult to imagine isn't occurring in any case of natural selection.

In this case, the answer to what sort of “cause” is natural selection is that it is a kind of relation in regards to prevention and omission that favours the successful reproduction of organisms with a greater degree of fitness. Their adaptive traits prevent death and omit infertility which allows for the inheritance of these traits within future generations (assuming the adaptive traits are inherited). Physical causation as described by CQ theory plays a role in this due to the innumerable physical processes and biochemical reactions that are governed by conservation laws that make such events possible. Everything biological is ontologically reducible to the entities described in the domain of chemistry, and the behaviour of atoms and molecules is dictated by the laws of thermodynamics and especially conservation laws of energy, and charge.

There are still however objections that need addressing. What about the example of smoking causing lung cancer? Sure we can state that carcinogens in cigarette smoke prevent the normal functioning of cells which results in cancer, but we also know this doesn’t always happen even sometimes in the most aggressive of smokers. There isn’t a universal relation between whenever a sufficient number of cigarettes are smoked and a sudden guaranteed presence of cancer. We may say that there is a causal relation however between smoking and the risk of cancer, i.e. smoking affects the probability of developing cancer even in the case where you do not develop cancer.

There is something to be said that on the microscopic level, there is a mechanism by which the contents of cigarette smoke via chemical and physical principles lead to the development of cancerous cells. Of course, these mechanisms do not happen in isolation, and various biochemical and individual factors act in the prevention of cancer forming, as well as immune responses that respond when cancerous cells do develop to prevent their spread. In terms of sheer frequency, the latter happens more often than a direct development of cancer from cigarette smoke, but this causal interaction between cigarette smoke and the cells of one’s lungs simply would not occur in those who are never exposed to cigarette smoke.

With a long enough time frame and frequent enough exposure, with slight variations to all the relevant variables eventually, cancer will develop – though not in a way so that the majority of smokers are certain to develop cancer. It just so happens that the frequency of lung cancer in the smoking population is several times higher than that of the non-smoking population and this is what we mean in regards to smoking effects on cancer risks. Regardless, there is still a causal mechanism to which cigarette smoking causes cancer, and this is due to a series of relevant physical and biochemical factors that are still dictated by principles restricted by the conservation of energy, charge, etc.

We can say in the context of omission and prevention as analyzed by Dowe (2007) that they explicitly include processes that are examples of physical causation as described by CQ theory. We can therefore claim that causation by prevention and omission supervene a complex series of physical causes that directly entails the exchange of conserved quantities.

Living populations in their environments are no exception, and it’s not something peculiar to acknowledge that the ecosystem and all the living creatures within it have their outcomes determined by the distribution and conservation of energy according to thermodynamic principles. This is revealed most readily through the study of trophic structures, digestion, and cellular respiration.

Natural selection however is not particularly about any of these particular physical processes as it’s an inherently multi-realizable event and necessarily has to be for evolution to occur in a variety of different environmental and temporal contexts.

Natural selection then would be thought of as a “cause” so much as it is a process of environmental prevention or omission that in turn affects an individual organism’s survival and/or reproduction, which then affects the frequency of certain alleles in the population as a whole. All of these supervening on a web of physical causes, which in turn involve the exchange of conserved quantities. When studying natural selection, we are studying macro-level phenomena of biological prevention and omission in terms of reproductive success and allele frequency.

As an example, we could look at the oxygen crisis that happened nearly two billion years ago. Before this event, the Earth’s atmosphere was mostly made up of carbon dioxide, nitrogen, and possibly methane. A massive increase in photosynthesizing cyanobacteria leads to an increasing abundance of molecular oxygen. Those with a background in chemistry know that the term oxidation is about the stripping of electrons from one atom to another due to oxygen’s noted tendency to take the valence electrons of other atoms. This sudden increase in atmospheric oxygen would lead to oxidation reactions in various living organisms, which would alter the molecular structures found within their cellular bodies. The changes in molecular structures were often fatal to these single cellular organisms resulting in their death. The ones that had structures that could maintain functioning despite changes, or didn’t react to oxygen allowed certain organisms to continue thriving.

In this case, we have an example of oxygen preventing regular cellular function within these single-cell organisms. If these structures acted normally the cell would continue to live. The structures behave as they do because of chemical principles that directly involve the conservation of energy and charge. In the event of an oxidation reaction, there is a clear exchange of charge with the exchange of electrons, and the total charge between reacting atoms is conserved.

The above example describes an instance of natural selection supervening physical causes described by CQ theory. The presence of oxygen prevented the continued reproduction of certain organisms, and the absence of structures negatively impacted by the presence of oxygen omitted the death of other organisms. This resulted in the outcome of living populations having an increased frequency of oxygen-tolerant organisms, all the while many of the organisms that couldn’t tolerate oxygen were removed from the gene pool.

Studying Causal Mechanisms in the Context of Natural Selection[edit]

An issue with the previous section is that Dowe (2007) did not commit himself to the belief that connected events through prevention and omission were genuine examples of causation. If this is the case and we classify natural selection as being a family of events characterized by prevention and omission then we would have to conclude that natural selection is not a true example of causation. Given that, what would it even mean to study the causal mechanisms of natural selection?

As said before natural selection supervenes on physical causes, this was exemplified in the example regarding the great oxygen crisis described above. There are very well causal mechanisms especially related to physical interactions and chemical reactions that can be directly studied that result in the macro-level phenomena of natural selection. This does not by any means require going above and beyond the traditional methods of testing adaptive hypotheses in the discipline of evolutionary biology.

Pigliucci & Kaplan (2006) reference various methods that evolutionary biologists use to test adaptive hypotheses on living populations; those being, phylogenetic analysis, optimization analysis, phenotypic manipulation, transplant studies, laboratory evolution, and regression analysis. It is noted that no method is truly sufficient on its own to provide indisputable evidence of any given adaptive hypothesis, but using multiple methods in tandem would provide the evidential grounds for considering a given hypothesis as being well supported (Pigliucci & Kaplan, 2006).

We can construct examples of certain methods and apply the above CQ theory extension of omission/prevention to demonstrate how the above methods work as genuine means to study the causal mechanisms of natural selection. Take as an example laboratory evolution.

Say for the sake of argument that you want to test antibiotic resistance in bacteria as a product of natural selection. You have two different samples of the same species of bacteria, one being your experimental group and the other being your control. Both groups will go through n generations within the experiment before being exposed to the same antibiotic. Before that the experimental group gets exposed to a weak dose of the antibiotic penicillin which is gradually increased in strength in every subsequent generation of bacteria if any happen to survive. The control group is left in a controlled environment free of any kind of antibiotic. Over time bacteria in the experimental group become increasingly more and more resistant to penicillin. After the final experimental generation, the two groups are then placed in an environment exposed to the same strong dose of penicillin. You observe a statistically significant effect where a drastically greater number of the bacterium from the experimental group survive the new exposure, while nearly all the bacterium in the control group dies. You have successfully observed evolution and natural selection in the lab, so how does our above theory of “causation” fit into this experiment?

The introduction of penicillin to the environment prevented the continued survival and reproduction of most of the bacteria that did not possess the trait of antibiotic resistance. The omission of penicillin in the control group allowed similar organisms lacking these traits to flourish normally. The mechanisms by which this happens can be explained biochemically and via other molecular mechanisms. This requires the transition or exchange of quantities like kinetic energy and charge which again obey conservation laws.

Phenotypic manipulation either through physical alteration of developed traits or altering a developing embryo allows scientists to identify how the presence or absence of a given trait can prevent or omit actions or events relevant to reproduction and survival. The same can be said of transplant studies.

This is not to say that the statistical methods criticized by Pigliucci & Kaplan (2006) are therefore useless in the event of actual experimental manipulation; such experiments often require the use of such statistical methods to make a quantifiable measure of the living population and how it is affected by certain environmental factors. Regardless as Pigliucci & Kaplan (2006) emphasize such statistical methods do not reveal the causal mechanisms surrounding natural selection, though that is not to discount them as giving clues.

To appeal back to the Indonesian shadow puppet analogy, sure studying the shadows of the puppets alone won’t reveal their structure or mechanisms – but using the shadows as a reference to recreate to build puppets of your own that match the shadow still provides considerable evidence to what those structures and mechanisms would likely be.

How we determine adaptation and likely- -selection experimentally gives rise to some of the methodological problems of evolutionary psychology as explored by Pigliucci & Kaplan (2006). Apologists will sometimes characterize critics of the discipline as attacking the theoretical premise of evolutionary psychology rather than its applications and methods. The premise that as a product of our evolutionary lineage the human brain would be shaped by past selection to give rise to behaviour or functions that would previously be advantageous is uncontroversial among serious evolutionary biologists and neuroscientists — trivial even. The problem is how the field of evolutionary psychology goes about testing adaptive hypotheses and its total disregard for sociocultural hypotheses for the same behaviour.

Pigliucci & Kaplan (2006) note that many of the methods used to test adaptive hypotheses such as phenotypic manipulation, transplant studies, etc. are either unethical or simply unfeasible to perform on human subjects. Many of the statistical methods mentioned earlier wouldn’t work well in human populations due to the very little genetic variation in comparison to other species.

This doesn’t make the study of human evolutionary adaptations hopeless, quite the contrary when sufficient variation and identifiable selection pressures are evident. Pigliucci & Kaplan (2006) talk of regions of the world where malaria is rampant and the heterozygous form of sickle-cell anemia provides resistance to malaria. Given that the direct genetic contributions, developmental pathways of the trait, and its responsiveness to the presence of malaria in the human population can be directly studied, a considerable degree of evidence can be provided for the trait being an adaptation. This only works because of how uncommon the trait is outside of malaria-ridden populations; it provides control populations in varying environments to see how the trait responds differently in different environments. For nearly universal traits, this isn’t as doable, which is why there are competing evolutionary hypotheses in fields like anthropology to how traits like bipedalism developed (Pigliucci & Kaplan, 2006).

Conclusion[edit]

Pigliucci & Kaplan (2006) criticisms of various statistical methods used within population genetics and evolutionary biology focus on the inability of such methods to investigate directly the causal mechanisms underlying evolution and especially natural selection. It had been noted that “causal mechanisms” was left unanalyzed and undefined and so the object of this essay was to provide such an analysis.

It was concluded that the causal mechanisms underlying natural selection were a matter of prevention and/or omission supervening on physical causes characterized by Phil Dowe’s CQ theory of causation. This account is consistent with the general Darwinian description of natural selection and fits with Pigliucci & Kaplan’s (2006) listed methods for testing adaptive hypotheses. Other theories of causation from the constant-conjunction theory, and the counterfactual theory had to be rejected for being too internally problematic to apply in the case of natural selection.

It has come to this author’s attention that there exists academic debate as to whether or not natural selection itself is even a causal process as opposed to being merely a statistical process — a formal inevitability. This paper would fall on the side of it being a “causal” process in so much that cigarette smoking causes cancer, or attentional neglect causes accidents is a “causal” process.

Knowing the kind of causation we are studying informs researchers on how to go about studying it. The kinds of causation we study in fields like physics and chemistry reduce to exchanges of conserved quantities which can be measured and modeled in incredibly controlled experiments. For the life and social sciences however we are studying a different kind of “causation”, one that is a matter of prevention/omission. This cannot be so precisely controlled or measured, and require more nuanced investigation. An investigation less like a physics experiment, but as Pigliucci & Kaplan(2006) say, more like a traditional detective mystery. In that, we have to be acutely aware of our limitations and we should not pretend to have a better claim to knowledge simply because the more rigorous methods and stronger justifications are less available. Lowering standards does not make our case stronger.

References[edit]

Collins, J. (2007). Counterfactuals, Causation, and Preemption. 10.1016/B978-044451541-4/50029-4

Dowe, P. (2007). Physical Causation (2nd ed.). Cambridge University Press.

Galavotti, M. C. (2018, July 13). Wesley Salmon (Stanford Encyclopedia of Philosophy). Stanford Encyclopedia of Philosophy. Retrieved May 6, 2023, from https://plato.stanford.edu/entries/wesley-salmon/

Hume, D. (2008). An Enquiry Concerning Human Understanding (P. F. Millican & P. Millican, Eds.). OUP Oxford.

Lewis, D. (1973, October 11). Causation. The Journal of Philosophy, 70(17), 556-567. http://www.jstor.org/stable/2025310.

Moore, D. S., & Shenk, D. (2016). The Heritability Fallacy. WIREs Cognitive Science. 10.1002/wcs.1400

Pigliucci, M., & Kaplan, J. (2006). Making Sense of Evolution: The Conceptual Foundations of Evolutionary Biology. The University of Chicago Press.

Psillos, S. (2002). Causation and Explanation. Acumen.

Salmon, W. C. (1994, June). Causality without Counterfactuals. Philosophy of Science, 61(2), 297-312. http://links.jstor.org/sici?sici=0031-8248%28199406%2961%3A2%3C297%3ACWC%3E2.0.CO%3B2-S