I was convinced that questioning my own beliefs is good and necessary while reading a book by Peter Boghossian and James Lindsay called, How to Have Impossible Conversations: A Very Practical Guide. The authors suggest an excellent way to examine and calibrate the confidence of one’s own beliefs is to determine the conditions that would need to be true in order to change that belief, also known as disconfirmation conditions. I decided to explore and possibly define disconfirmation conditions for my own religious beliefs. The entire experience was fascinating and as I recount the details here, I hope you find it as interesting and thought provoking as I did. In this article, I present a summary of my resulting disconfirmation conditions, and the first of a two-part analysis of the plausibility of those conditions.
Summary of Disconfirmation Conditions
My belief in the existence of a creator would be disconfirmed if one could demonstrate experimentally that macroevolution can occur between two animals, such that:
It is possible to complete the experiment within the span of a human lifetime.
The two animals are sufficiently different from each other such that the transition is clearly distinguishable from examples of micro-evolution. An example of macro-evolution would include:
major morphological changes
viable changes to sexual organs
viable changes to vital organs
An example scenario that I would find convincing is an experiment in which a fruit fly is mutated over multiple generations until the result of the last mutation is a flea.
Plausibility Analysis
To construct an experiment to recapitulate the evolution between two animal descendants and their common ancestor is a notoriously difficult problem to solve. Two primary challenges offer the greatest obstacle to fulfilling the plausibility requirement for my disconfirmation conditions, both related to time. The first challenge is how long it takes for macroevolution to occur and the second challenge arises from the time it takes to solve certain types of problems.
In his book, Why Evolution is True, Jerry Coyne eloquently describes the first challenge of observing or developing an experiment to reproduce macroevolution:
Further, we shouldn’t expect to see more than small changes in one or a few features of a species—what is known as microevolutionary change. Given the gradual pace of evolution, it’s unreasonable to expect to see selection transforming one “type” of plant or animal into another so-called macroevolution within a human lifetime. Though macroevolution is occurring today, we simply won’t be around long enough to see it.
-- Jerry Coyne, Why Evolution Is True
Coyne clearly highlights the primary challenge to the plausibility of the experiment my disconfirmation conditions require. Will there be enough time to complete such an experiment? Can an experiment be crafted to reduce the time required for macroevolution to within a human lifetime? I suggest that the time required to perform such an experiment can be reduced significantly using a two-fold strategy.
The first part of the strategy to reduce the duration of the experiment comes from recognizing that much of the time required for macroevolution to take place is consumed waiting for the next mutation in the required sequence. A modern experiment can use gene-editing technology, to direct the mutations of an animal, removing the need to wait for the next mutation to be the right one and dramatically reduces the time for my target sequence of mutations to occur.
The second part of the strategy to increase the feasibility of the experiment, is choosing two animals with short life cycles, ideally with very little time between birth and reaching sexual maturity, like many insects. How can I determine if this strategy reduces the duration required for macroevolution into a single human lifetime?
To try to quantify the problem, I will have to make some assumptions, so I will try to err on the conservative side when making those assumptions. The equation I am using to approximate the time required for the experiment is:
Experiment duration = (Required mutations) * (Avg period between required mutations + Avg period from birth to sexual maturity)
If I can force the next required mutation to take place each generation, this reduces the “Avg period between required mutations” to 0, reducing the experiment duration to the product of the number of required mutations and the average period from birth to sexual maturity for the flea and fruit fly.
Experiment duration = (Required mutations) * (Avg period from birth to sexual maturity)
Starting from the hatching of the egg, and averaging the period for each stage of life leading to adulthood, the generation period can be determined for both insects. The resulting average generation period can then be applied to the equation for estimating the duration of the experiment.
Generation of flea = (5-10 days for larva) + (8-9 days for pupa) = (5+15)/2 + (8+9)/2 = 18.5 days
http://bioweb.uwlax.edu/bio203/s2008/kwong_hoi/lifecycle.htm
Generation of fruit fly = (9-10 from egg to adult) = 9.5 days
https://depts.washington.edu/cberglab/wordpress/outreach/an-introduction-to-fruit-flies/
Average generation period = (18.5 + 9.5)/2 = 14 days
Experiment duration = (Required mutations) * 14 days
Notice the last remaining value that must be obtained to solve the equation is the total number of “Required mutations” to perform the desired transition between animals. Is it possible, based on our current knowledge, to derive a reasonable estimation of this number? Before attempting to obtain a good estimate for the number of mutations required, let’s examine a few hypothetical numbers using the current equation. Assuming the number of required mutations is 10,000 would yield:
Experiment duration = 10,000 * 14 days = 140,000 days = 383.56 years
The resulting duration is 4-5 times the length of the average modern human, so additional optimization of the experiment is required to further reduce the overall duration of the experiment. However, I am already surprisingly close to a feasible experiment if the number of required mutations is anywhere close to 10,000. To put the success so far into perspective, according to the study I am using for my thought experiment, the common ancestor for both the fruit fly and the flea originated about 250 million years ago, with the flea and fruit fly originating about 85 mya.
Image annotated from original posted by sci.news https://www.sci.news/biology/science-family-tree-insects-02264.html
So, the combined period of evolution between the two animals and their common ancestor is:
Period of evolution = (250 – 85) * 2 = 330 million years
This underlines the power of forcing the desired mutation each generation, with the overall experiment duration being reduced from 330 million years to 384 years, assuming the 10,000 number for required mutations is an approximation of the actual number. There is no need to attempt to reduce the duration of the experiment further before having some way to measure the accuracy of the overall number of required mutations, which brings me to the analysis of the problem of determining this number, and strategies to overcome this problem. So let’s bracket the time required to complete the experiment for a moment and explore the second primary time challenge of constructing this experiment.
In order for this experiment to be successful, the scientist must know not only the correct mutations, but also the entire correct sequence of mutations between each animal and its common ancestor in order to use gene-editing technology to direct the next mutation in the required sequence each generation. To obtain this sequence of mutations, the scientist must seemingly solve a problem that approximates the shortest-path problem, which is disconcerting given the extremely large dataset. In other words, it would be unreasonable to expect to solve such a problem through brute-force calculations within a human lifetime using modern computing power. In order to obtain a viable mutation path to use in my experiment, I will need to employ several strategies to dramatically and accurately reduce the dataset that requires searching, which I will examine in my next article.
Thank you for taking the time to read about my experience determining the disconfirmation conditions for my own religious beliefs. If hope you found this interesting, I encourage you to keep an eye out for my second and last article on plausibility in this series, followed by an article describing why these conditions merit disconfirming my religious beliefs, and a final article reflecting on several challenges I encountered during the process. I am eager to hear any comments or questions you might have about the disconfirmation conditions, the plausibility of the experiment, or any part of the experience.