As a professor, one of the highlights of my role is participating on committees of PhD candidates. The four-to-six-year process of earning a PhD is an intellectually challenging and often psychologically taxing process which we as faculty are mentoring students to help them clear the academic and personal hurdles that many of them face on this journey. By the end of the second-year, prospective PhD candidates will need to pass a comprehensive written and oral examination given by a committee of five PhD faculty in order to continue in the program.
I recently participated as a committee member in a student’s oral examination. This one was particularly engaging as we navigated the complex terrain of historical versus experimental science. We explored foundational questions about the origins of life and the very definition of life itself. Two hours was not nearly enough to satisfy a group of faculty which included a geochemist, a philosopher, an analytical chemist, a biogeochemist, and a biologist (that would be me) that all had many questions and wanted to engage in the discussion. This was true especially as we veered into topics such as how we might identify life on Mars (or other planets and moons) or assess the probability that life once existed, which then necessitates a discussion of how one defines life at all.
This oral exam follows several weeks after the student’s comprehensive written exam. The two combined provide a pivotal moment—both for the student and the examining committee. It’s the committee’s opportunity to delve deeply into the nuances of scientific inquiry, challenging the student with thought-provoking questions that span the broad spectrum of science, from philosophical underpinnings to practical methodologies. This process is crucial as it prepares the student for the demanding task of scientific research and scholarship. The committee will either affirm that the student is capable of taking on and completing a PhD research project and thus allowed to advance to the next steps of the program, or may find the student unlikely to be able to complete the entire process, in which case they are encouraged or required to leave the program.
Today, I’m excited to share, with permission, the written examination responses from a student (RR) for whom I am a committee member and who has just passed her oral examination.
Each member of our committee contributes to shaping these exams, providing specific reading materials and questions that challenge the students to think critically and articulate their understanding.
Here, I’m going to share the reading list that I provided RR many months ago and my directions. Then I’ll present the prompt I gave to her, including directions for my 6-hour exam time and then I will share her insightful responses.
The essays you will read below are responses to prompts I provided, designed to test RR’s ability to synthesize and critique the information she has studied. These essays not only reflect the rigorous academic standards we uphold, but also encapsulate RR’s ability to bridge complex ideas across historical and experimental sciences. They address central issues about the reliability and interpretation of scientific evidence, a topic that resonates deeply with ongoing debates in science, such as those posed by proponents of creation science who challenge the validity of historical sciences.
I’m sharing these because I thought my blog and YouTube followers might be curious about the process that students (and myself at one time) go through in their training to get a PhD. I also thought these were especially good responses and would be interesting to many people who follow my blog. I’ve talked about many of these issues here on my blog and on my YouTube channel especially with respect to the validity and nature of historical versus experimental sciences. I’ve referenced Dr. Carol Cleland numerous times and I challenged RR to read her work and to critique it. You’ll see the material I gave the student involved a lot of her writings. I think that RR’s work deserves to be seen by more than just 5 PhDs sitting in their office cubicles!
In her essays, she presents a nice concise summary (note that she is under a severe time constraint of 6 total hours to compose two essays, plus read a new paper and integrate it into her writing) of how historical and experimental sciences approach evidence, a central question about how we have confidence in what we know about things that happened in the distant past even though we weren’t there to see it. This connects with my interest in creation science apologists such as Ken Ham from Answering in Genesis who constantly preaches that historical science isn’t a trustworthy exercise and should be doubted because only eyewitness testimony or actual things that can be observed in the present (something he calls “observational science”) can really be trusted. This whole question of the validity of historical sciences is therefore an important one in circles that I am engaged.
A bit of background before we get to the comprehensive exam. This particular student, RR, is involved in origins of life research and is interested in things that potentially happened over 4 billion years ago. I don’t have time to go into her whole research program—it’s quite fascinating, including the specifics of biochemistry research her lab is actively researching—but I thought it was important for her to think about historical science and the nature of historical sciences since she is deeply involved in a historical science. That is why I gave her this particular material to read.
So, without further ado, here are my prompts and her responses. With respect to her responses. These are not verbatim from her exam but have been copyedited lightly after her submission.
RR Comprehensive Exam Materials from Joel Duff
Hi RR, I am providing a set of research publications and popular overview articles as reading material. My goal is to challenge you in areas outside of your area of expertise. I want to challenge you to think in terms of big questions in biology and science in general.
The two areas that these resources will help you to understand better are:
- Philosophy of science: Historical vs experimental sciences
- Origins of Life research just focused on issues around thermodynamics and a definition of life
You should not be reading these papers trying to memorize every detail. I’m looking at how you synthesize the big pictures from these papers.
Reading material on the philosophy of science (historical vs experimental science)
Cleland, Carol E. “Methodological and epistemic differences between historical science and experimental science.” Philosophy of science 69, no. 3 (2002): 474-496.
Cleland, Carol E. “Prediction and explanation in historical natural science.” The British Journal for the Philosophy of Science (2011).
Note: I only gave her papers from a single philosopher but I did so knowing that she is fairly well read in philosophy of science having been assigned many chapters from another faculty member on origins of life questions and issues. I also knew that I was going to challenge this student with an additional reading on the day of her written exam that would be critical of Cleland’s work and thus I wanted to see how she would integrate that new information in a short space of time.
Reading material for the origin of life
Cleland, Carol E. “Life without definitions.” Synthese 185 (2012): 125-144.
Michaelian, Karo. “Non-equilibrium thermodynamic foundations of the origin of life.” Foundations 2, no. 1 (2022): 308-337. https://www.mdpi.com/2673-9321/2/1/22
Book: “Every Life is on Fire: How thermodynamics explains the origins of living things” by Jeremy England. I will give you my copy to read.
Duff Written Comprehensive Questions for RR
Time allotment: 6 hours. Start time: 10:40 am; end time: 4:40 pm
Conditions: You may have and reference all articles that I provided to you previously and any other papers you might have found that helped you prepare. You should compose your answers on a computer but no internet sources should be available. I have provided additional material which you will need to read and will play a large part in the assignment below.
Assignment: You will response to two writing prompts. By doing so you will demonstrate your knowledge of concepts, your use of disciplinary specific language, and your ability to think critically and apply important theories to your research area.
First Prompt: Compose an essay or narrative in which you compare, contrast and respond to the following article that I will provide for you to read during your time dedicated to responding to my questions: Turner, Derek. “Local underdetermination in historical science.” Philosophy of Science 72, no. 1 (2005): 209-230. Then contrast and compare Turner’s contribution to the question to the proposals of Carol Cleland that you have already read. The response would be your own assessment of these approaches to validating or determining the strength of conclusions of historical science. Some reflection on how these ideas may influence or improve your understanding of the research you are participating would also be appreciated.
RR’s Response:
Cleland makes the case that historical science and experimental science utilize fundamental methodological differences for the testing of their hypotheses, which represent different patterns of evidential reasoning. Experimentalists in the classical sense, investigating regularities among event-types, are principally focused on evaluating the validity of repeatable generalizations; they generate predictions from a single hypothesis and manipulate repeatable test conditions in the lab while controlling for factors that could produce false positives or false negatives (Cleland 2002). Historical researchers in the prototypical sense, investigating event-tokens, are principally focused on evaluating mutually exclusive hypotheses about specific past events and hunting for present-day evidentiary traces to distinguish among them (Cleland 2002). These events cannot be reproduced in the lab, but historical researchers can look for present-day traces of the events, which culminates in the search for a ‘smoking gun’: the trace(s) that unambiguously highlights one hypothesis as providing the best explanation based on the available evidence (Cleland 2002). These two patterns, based on different types of evidence available to the historical researcher vs the experimental researcher, are employed in order to deal with different sides of an asymmetry of epistemic overdetermination (Cleland 2002). These differences, however, do not support the common claim that historical science is epistemically inferior to experimental science; on the contrary, Cleland argues that historical science and experimental science are, in an epistemic sense, equivalent (Cleland 2002). Put succinctly by Turner, her argument is as follows:
P1. Later affairs usually overdetermine earlier affairs, but earlier affairs usually underdetermine later affairs.
P2. Historical scientists exploit one half of this asymmetry: their methods for testing hypotheses about past event tokens are appropriate because later affairs usually overdetermine earlier affairs.
P3. The experimental method is a strategy for coping with the other half of this asymmetry: since earlier affairs and events (such as a short circuit) usually underdetermine later affairs and events (such as a house burning down), anyone who wishes to test hypotheses about regularities among event types must run a series of trials in which different test conditions are manipulated, with the aim of ruling out false positives and false negatives.
C. Therefore, prototypical historical science and classical experimental science are equally good, epistemically speaking. (Turner 2005, p. 213)
Turner has two problems with Cleland’s claim that historical science and experimental science are equally good in an epistemic sense. His first problem lies in the difference between the causal/metaphysical overdetermination invoked by Lewis (the original proponent of the theory of asymmetry of overdetermination) and the epistemic overdetermination Cleland seems to be invoking (Cleland 2002, Turner 2005). According to Turner, Cleland is “misled into thinking the conclusion follows from the premises because she fails to distinguish clearly between causal/metaphysical overdetermination and epistemic overdetermination” (Turner 2005, p. 213). Consequently, Turner argues that causal/metaphysical overdetermination does not imply epistemic overdetermination—as Cleland claims—but rather epistemic underdetermination (Turner 2005). To make his point, he uses one of Cleland’s own examples (which she used to demonstrate the thesis of asymmetry of overdetermination): the case of a baseball shattering a kitchen window. The baseball shattering the window does not overdetermine the later traces (such as the shards of glass landing on the floor), but in a metaphysical/causal sense of overdetermination, there are many subcollections of traces that overdetermine the baseball’s hitting the glass (Turner 2005). One does not need every single trace on hand in order to reconstruct what happened; any number of traces—the shards of glass, the baseball on the floor, the size of the hole in the window—would be sufficient to conclude that the event happened.
But Turner, (cleverly, in my opinion), points out that this is not the situation in which many historical researchers find themselves. The situation more analogous to that encountered in historical research is if we imagine the owners of the house clean up the mess before the researcher arrives. They store the baseball away in the garage, sweep up the shards of glass, and replace the window, so that in a few weeks (or decades)—when the historical researcher arrives—the only traces of the event that remain are a few shards of glass shoved underneath the refrigerator, gathering dust (Turner 2005). The house cleaning and repair are examples of information-destroying processes, which occur constantly in nature. The historical researcher who has stumbled upon the dusty shards of glass is now faced with an epistemically different situation. Based on the only available evidence (the shards of glass), they cannot at all discriminate between incompatible rival hypotheses (window vs. wine glass, home break-in vs. flying projectile, football vs. baseball vs. bird, etc.). To make matters worse, because the researcher knows that people tend to clean up when things happen that cause shards of glass to go flying everywhere, they have good reason to think that the few dusty shards of glass represent the only evidence that will ever be available to them, thus rendering any chance of finding a smoking gun and distinguishing between rival hypotheses hopeless (Turner 2005). This situation represents a local epistemic underdetermination problem, which is clearly not ruled out by causal/metaphysical overdetermination. Because earlier events can be epistemically underdetermined by their later traces—according to Turner—the time asymmetry of overdetermination thesis does not lend itself to Cleland’s thesis regarding the relative epistemic status of historical vs. experimental science (Turner 2005).
Turner’s second problem with Cleland’s claim of epistemic equivalence between historical and experimental science is that local epistemic underdetermination is more pervasive in historical science than experimental science, largely due to the roles that background theories play in each type of research (Turner 2005). In historical science, background theories tell us that many historical processes such as fossilization, weathering/erosion, continental drift, subduction, glaciation, etc., are information-destroying processes just like the house cleanup or window repair in the baseball-shattering-window example (Turner 2005). In many cases (I suppose Turner would argue in almost all cases), background theories in historical science give us good reason to think that the current traces represent all the empirical evidence that will ever be available to us. This means that a smoking gun may never be found, rival hypotheses would be empirically equivalent in the strong sense, and thus the conditions for local epistemic underdetermination would be met (Turner 2005). While historical researchers can use experimental science techniques to develop new technologies for identifying and studying potential smoking guns, the fact remains that they can never create a smoking gun; if one existed, but nature destroyed it, there is nothing to be done about it (Turner 2005). Turner argues that this role of background theories in historical science should therefore limit the epistemic ambitions of historical scientists.
The role of background theories in experimental science are different, resulting in local epistemic underdetermination being less of a problem. Here, background theories can expand the epistemic ambitions of experimentalists (Turner 2005). An experimentalist can use background theories to build new apparatus or manipulate test conditions; they can also use background theories to inform the design of new experiments whose purpose is to produce results that break the tie between weakly empirically equivalent hypotheses (Turner 2005). In other words, new phenomena, or new evidence, can be produced in the lab, which is not possible in historical science. Ultimately, the different functions of background theories in historical science compared to experimental science result in local epistemic underdetermination being a more widespread problem in historical science than experimental science, thus providing one sense where historical science is epistemically inferior to experimental science, and thus falsifying Cleland’s conclusion that the two are epistemically equally good (Cleland 2002, Turner 2005).
I think that Cleland’s overarching point is that since experimental and historical scientists exploit different aspects of the asymmetry of overdetermination, that neither practice may be held up as more objective or rational than the other (Cleland 2002). As Turner points out himself, his claim that historical science is, by way of the pervasiveness of local epistemic underdetermination, epistemically inferior to experimental science does not void Cleland’s point. I think this is a justified view—one I agree with—and it highlights that the conclusions drawn from historical science are in no way weaker than those drawn from experimental science. However, I disagree with Turner’s narrow, limited role for background theories in historical science. Historical researchers can use experimental science techniques to expand the range of observables, thus pushing rival hypotheses from being strongly empirically equivalent to only being weakly empirically equivalent and allowing for new techniques that break evidentiary ties among weakly empirically equivalent hypotheses. Importantly, Cleland (2011) distinguishes between information-destroying processes and information-degrading processes, arguing that actual information-destroying processes are much less pervasive than Turner believes. Indeed, scientists have become increasingly better at expanding the range of observables.
In origins of life research, experimental work is often used to investigate a hypothesis bearing a tenuous logical relation to the target hypotheses (Cleland 2002). An example is the classic Miller-Urey experiment, which set out to test the hypothesis that life originated in a primordial soup but was really a test of the idea that some of life’s important building blocks could be produced by a spark-discharge experiment under atmospheric conditions believed to be prebiotically relevant. I do not believe a smoking gun that unambiguously clinches the case for a particular origins of life hypothesis—for example, metabolism-first or replicator-first—will ever be found because of the ‘information-destroying processes’ of chemical and biological evolution which have erased chemical fossils. Cleland’s distinction between information-destroying processes and information-degrading processes is irrelevant in the respect of chemical fossils. Turner states that “historical scientists are trained to identify local underdetermination problems and to move on to more tractable research questions…for this reason it would be difficult to produce examples of research problems that scientists are currently working on, and that clearly satisfy the conditions for a local underdetermination problem” (Turner 2005). Origins of life research is an example of just the case he described; however, origins of life researchers, I think, are looking not for a smoking gun to differentiate between strongly empirically equivalent hypotheses, but rather for hypotheses which are the most unifying and that have the best explanatory power (common cause explanations).
Origins of life research is a perfect example of the role that the narrative and common cause explanations can play. In a narrative explanation, “the basic idea is to contrast a story—a coherent, intuitively continuous, causal sequence of events centering on a precipitating event and culminating in the traces of an explanation” (Cleland 2011, p. 567). Sometimes, the purpose of the narrative is to simply establish that certain sorts of causal processes could have plausibly led to the phenomena concerned, and sometimes the purpose is to show how the phenomena actually happened (Cleland 2011). In a common cause explanation, “the basic idea is to formulate reliable inferential methods for identifying when a diversity of contemporary traces comprises the effects of a common cause token” (Cleland 2011, p. 567). Often times, narrative and common cause explanations are used in conjunction with each other, where common cause explanations provide the needed empirical warrant for critical events in the narrative sequence (Cleland 2011). A common cause hypothesis that explains the total body of evidence better than any of its scientifically plausible rivals is therefore most likely to be true (Cleland 2011). I think origins of life hypotheses can be judged on the basis of common cause explanations, but at the same time, origins of life researchers should broadly accept that we will never truly know how life originated on Earth, and subsequently refrain from making any claim that one hypothesis is the definitively correct one over another. The few researchers in the field who do attempt to make such claims, in my view, are incorrect to do so.
Second Prompt: Thinking about the origin of life and life itself here are three questions to respond to: 1) What was your reaction to Jeremy England’s dissipation-driven adaptation model for the origin of life? In your answer you should be sure to describe his thesis in layman’s terms. 2) Do any of his ideas have any impact on the processes that you have been working on and if so how? 3) After reading some about definition of “life” at this point what would be your definition of life and your defense of it? While defending a particular definition you may also wish to point out some of the weaknesses or unanswered questions.
RR response:
England devotes his book to building the conceptual framework needed to lay out an argument couched in physical terms for a mechanism of emergent lifelike behavior that is not founded on self-replication and natural selection. He proposes that the field of nonequilibrium thermodynamics is starting to show us how to break the stepwise process of life’s emergence into comprehensible increments that seem independent of metabolism-first, replicator-first, etc. hypotheses of the origins of life. The mechanism is that of dissipative adaptation, which, put simply, states that it is the relationship between how energy flows and how the medium (structure) it flows through becomes transformed by that explains the emergence of structures that recapitulate numerous different behaviors that we associate with distinctive qualities of a living organism (reproduction, harvesting energy from the environment, maintenance/repair, responding to stimuli) (England 2020). Once this cycle of influence goes through multiple iterations, a selection principle may come into play, allowing the structures and shapes of things to skew strikingly in a direction that reflects the relationship between shape and energy flow in a system (England 2020). This produces an analogous result to that observed in neo-Darwinian evolution, whereby genetic mutations lead to adaptations, which lead to natural selection. My main takeaway from England’s book is that the language of Darwinian evolution through natural selection can be translated to the language of thermodynamics, where both explanations are doing the same thing; on this view, Darwinian evolution is quite clearly an instance of dissipative adaption. Still, thinking about dissipative adaptation—especially in systems not yet fully understood using the language of Darwinism—may offer a new kind of opportunity to identify something going on.
In some ways, I view England’s dissipative-adaptation model as reminiscent of the physicalist/mechanist/reductionist models for the origins of life that dominated science prior to the advancement of the fields of biochemistry and molecular biology. This view held that life in general can be understood completely in terms of physics and chemistry; therefore, mechanists/physicalists/reductionists approached the problem of understanding life by searching for a more thorough explanation in purely physicochemical terms (Oparin 1964). The downfall of this view was triggered by its proponents’ insistence that there was nothing special about living things; indeed, the workings of living things were likened to that of machines. The holist or organicist paradigm that bolsters contemporary origins of life research accepts that molecular processes can be explained in terms of physics and chemistry, but that physicochemical mechanisms play increasingly smaller roles at higher levels of organization, at which time they are replaced or supplemented by biological laws describing emergent characteristics or properties of organized systems (Mayr, 1997). England’s dissipative-adaptation theory offers what appears to be another paradigm where physicochemical mechanisms maintain important roles even in increasingly higher-organized systems.
England’s idea of thinking about dissipative adaptation, especially in systems not fully understood using the language of Darwinism, is one I can use in my research regarding the transition from biomonomers (amino acids, nucleotide monophosphates) to their respective biopolymers (peptides, oligomers). I am very interested in the isomeric and chiral selection of biochemical universals (alpha-L-amino acids and D-sugars) and how this process could have come about through the coevolution of biopolymers; this would be a form of chemical evolution that preceded Darwinian evolution. Dissipative adaptation is the idea that “a system made of many pieces accesses energy from an external source differently depending on how those pieces are arranged, but that the energy that is absorbed also has an impact on how the arrangement of pieces can change over time” (England 2020). Michaelian’s thermodynamic dissipation theory of the origins and evolution of life suggests that the fundamental molecules of life—nucleic acids, nucleic acid polymers, amino acids, and fatty acids—were, at life’s origin, self-organized dissipative structures that originated and persisted for the fundamental physical purpose of producing entropy. The selection of these molecules, therefore, was because certain structures (including isomeric and chiral structure) did a better job at dissipating photons than the alternatives (Michaelian 2022). I think these ideas are very interesting to consider for my own work, especially if synergistic or mutualistic interactions between certain precursor molecules provided greater dissipative efficiency compared to the molecules working independently.
Dozens of attempts have been made to either define life or to list characteristics which are believed to be necessary and sufficient to life. However, all of them face the same problem: they are based on a single example of life—life on Earth. This poses a serious empirical problem in the sense that it is impossible to “safely generalize to all of life, wherever and whenever it may be found, from a single, potentially unrepresentative example of life” (Cleland 2012). Because extant life represents a single example in that it shares a last universal common ancestor, it is extremely difficult to discern whether ‘universal’ biochemical characteristics of life on Earth are universal because they are truly required for life, or if they may instead reflect chemical and physical contingencies present on the early Earth (Cleland 2012). I think it is imperative that we do our best to not be blinded by bias of what life looks like on Earth, because we risk missing alternative forms of life. I am in favor of a list based on the physical limits for life—which are likely to be approximately the same anywhere in the universe—that include only absolute requirements that would reasonably apply to life anywhere in the universe, regardless of specific chemical structure. The two most absolute requirements, listed by Benner et al. (2004) are (1) thermodynamic disequilibrium, which is needed for life to “do something”, and (2) bonding, which must support information transfer needed for Darwinian evolution (a combination of covalent bonding, hydrogen bonding, and the hydrophobic effect would all meet this requirement; in different solvents or at different temperatures, various bonding may be suitable) (Benner et al. 2004). Regarding the search for life elsewhere, I think Cleland is right to propose using “tentative criteria”, which should include a diversity of common and uncommon features of extant life since we do not know which characteristics are truly essential for all life vs. which characteristics simply reflect the physical and chemical contingencies of the early Earth (Cleland 2012). Importantly, these tentative criteria do not compete with each other and can be simultaneously used in the search for novel life. Additionally, they are constructed as tentative, and therefore understood to be incomplete, so that they can be adjusted depending on changes in our epistemic situation (Cleland 2012).
The question of “what is life?” has many moral, legal, and philosophical implications, and so broadly, I understand the general human desire for an answer to this question. It also clearly has implications for artificial and synthetic life. But for the purposes of searching for life elsewhere—which is commonly invoked as a reason for needing a definition—I do not see why we need a concrete, infallible definition. I think we should treat the problem of defining life on Earth as something separate and distinct from the problem of defining life for the search for it elsewhere. We could use a definition based on our single example of life, understanding that it applies only to life on Earth and may very well not be universal to all life, to address the moral, legal, and philosophical questions. Then, we could employ Cleland’s call for tentative criteria for the search for life elsewhere.
Naturalis Historia 
Comments or Questions?