Science and non-repeatable events

Scientific method is fundamentally concerned with repeatable events. The phenomena that science captures most easily may be described using the following formula: once conditions A have been established, if B is done, then C happens. 

This kind of science is a science of reactions, of the reactive. But what about a science of the active? Is such a science possible?

To phrase what I have in mind in a different way, suppose that there are events in our universe that are not reproducible or repeatable. They would not be the consequence of some stimulus or trigger. But neither would they be the act of some imaginary god. They might simply be part of the same underlying, mysterious generator that is responsible for what we call scientific laws (patterns of reproducibility). (So far we have inferred some of the properties of this generator, but we are very far from apprehending it or understanding its totality and boundaries. Intellectual humility is crucial.) Would science be able to record and theorise about such events? Certainly not. Modern scientific method is firmly aimed at eliminating irreproducible results.

To put it in still another way, we are able to verify determinism in those cases where it holds up, but we are always unable to verify the absence of cases (in the past or in the future) where the deterministic rules break down.

This is a quandary, since it does seem that the world contains phenomena that are difficult to reproduce. The belief that the world can ultimately be reduced to a set of deterministic rules is not at all uncontroversial (and perhaps many physicists have given it up already). Particularly in biology, we constantly struggle to understand phenomena in terms of such rules. However, we can perhaps see biology residing at the boundary between the reactive/deterministic and the active/irreproducible. Gradual determinism? —

Category: Bioinformatics, Computer science, Philosophy | Tags: , , , , , 5 comments »

5 Responses to “Science and non-repeatable events”

  1. Christian

    what about probability theory, or quantum mechanics (probability amplitude)?

    maybe they describe process outcomes because we don’t understand the process and how it is influenced?

  2. admin

    Every useful notion of common probability that I have seen seems to refer back to repeatability. For example, a distribution of people’s heights assumes that we can repeatedly pick people from some collection.

    I don’t know quantum mechanics as well as I should, but the notion of probability amplitude seems to also refer back to repeatability: if you observe the system sufficiently many times (thus “repeating” your observation from some given starting conditions, presumed to be identical!), eventually you should see the curve you expected.

  3. admin

    Still, I suppose that distributions offer a way out on some level: in a continuous distribution you can keep picking new values indefinitely without ever repeating a past result. Maybe the essence of a continuous distribution is this combination of an inexhaustible possibility of uniqueness (in new values) and predictability (in the overall curve)? But this is somehow not good enough, since you can always encounter “black swan” values that will break the curve (and you will do so in sufficiently interesting systems).

  4. Shandar Ahmad

    Seems an interesting thought. I am reminded of probability (as Christian speaks of) and I agree with Johan’s response that probabilistic reproducibility is repeatability after-all. Quantum mechanics works the same way too. Yet, Johan, a mere technical difficulty in assembling a population does not mean the theory breaks down. My greater concern is that the probability assumes some kind of distribution, which is often unknown, making it hard to assess the repeatability of an event. To the main point of your discussion, I guess dependence on some kind of repeatability is inevitable in science, because if it does not hold, the number of alternative hypotheses to examine become too huge- again a matter of convenience though. It would be interesting to examine the analytic tools we may want to use, if we go beyond this convenience of repeatability. And to think if this analysis will produce any tangible outcomes?

    Single-cell analysis of gene expression is picking up in biology and this discussion would have great relevance for the same.

  5. Johan Nystrom

    Shandar, thanks for your comment. Interesting to consider single-cell gene expression analysis in this context. I can imagine that repeatability would be very difficult there, even under the best experimental conditions.

Leave a Reply

Back to top