What is Science?

Hello everyone!


Due to my inability to use WordPress correctly, I’ve accidentally been saving my posts as drafts rather than posting them! Due to that, I’ve put up two posts back to back today and will try to be more careful in the future.


I’m now in my 5th week of classes, though it often still feels like I’ve just started.  For example, what I’d initially thought to be an interesting albeit trivial question, ‘what is science’, has in fact become more complicated the more I learn about it.

I’d been a bit taken aback my first week of classes when I learned that the typically accepted definition of science (by ‘accepted’ I mean the definition of science taught in elementary education – in University level science courses the question of what does and doesn’t constitute science rarely comes up other than to debunk poor research techniques) of observation-based conclusions used to prove and disprove theories is not generally accepted as ‘science’ from an epistemic point of view.  This topic had been debated heavily in my Philosophy of Science class – as one might expect.


We first looked at the Popperian view of science, which introduced the concept of falsifiability.  Popper says that an idea is more scientific the more falsifiable it is, the more predictive power it has, and the riskier it is.  Falsifiability, in this sense, concerns a concept’s capacity to be tested. So a theory such as intelligent design is not very scientific because one cannot test the claim that an intelligent being designed the world – this does not prove that intelligent design is a ‘false’ theory, just that it is not a scientific one by this standard.  Additionally, by this definition theories are never proven true, but are either falsified through testing or ‘survive’ for another round.  Predictive power is another necessity of a scientific theory.  A theory that merely explains events of the past (there have been unusual levels of flooding along the Louisiana coast for the past 10 years) is unscientific, while a theory that hypothesizes future events based on the past is more scientific (recent trends in flooding along the Louisiana coast suggest that more powerful storms are to come).  And where predictability is concerned, Popper says that the more specific a prediction is (and thus the more easily falsifiable), the more scientific it is.

My problem here is that I find it difficult to dismiss all conception of truth. While I don’t believe absolute truth is necessarily obtainable, I certainly think one theory can be more true than another, rather than just not-yet-proven-false.


We next looked at Kuhn’s interpretation of scientific development as it concerns paradigms. Rather than that science is a universal pursuit resulting in progressive accumulating knowledge, Kuhn asserts that science evolves in a messy fashion characterized by periods of ‘normal science’ and ‘revolutionary science’. A paradigm consisting of a central theory, auxiliary hypotheses, heuristic models, and various methods is developed during period of revolutionary science.  From this point, the paradigm grows as its auxiliary hypotheses are adapted and developed during periods of normal science where little questioning of the paradigm occurs.  Within this period, anomalies, or problems that the paradigm cannot solve, are discovered.  Given time a secondary paradigm will develop, under a new period of revolutionary science, which will take the place of the current paradigm (a paradigm shift) – here science leaps forwards, carrying with it some knowledge from the previous paradigm but leaving behind ideas that are incommensurable.  A paradigm is said to be ‘better’ than another if it solves problems more accurately, more consistently, with a broader scope, and/or with a simpler theory.

A typical example here is the switch from the Ptolemaic theory of planetary motion where the planets revolve around the Earth to the Copernican theory of planetary motion where the planets (including the Earth) revolve around the Sun. Ptolemy’s theory had been adapted during periods of normal science to continue to fit the growing knowledge of planetary motion, such that now the planets revolved on an axis which revolved around a location adjacent to Earth (i.e. Ptolemy’s theory was becoming excessively complicated in order to account for inconsistencies between theoretical and observed motion).  According to Kuhn, Copernican theory not only better explained planetary motion, but did so with simpler reasoning.


Just this last week we started looking at Lakatos’ concept of scientific research programs, an idea that attempts to find common ground between Popper and Kuhn’s ideas. Lakatos starts by distinguishing refutation from rejection.  He states that while particular evidence may refute a theory, by no means does this necessitate that the theory as a whole be rejected, as a naïve conception of Popper might suggest.  This, he argues, would lead to the rejection of all theories.  Thus, a systematic method of determining at what point refutation should lead to rejection must be developed, and, in the case of rival theories, it must be determined how one can determine which of a pair of refuted theories should be accepted, and which should be rejected.

Here Lakatos puts forth the concept of a research program, which he distinguishes from a paradigm. A research program consists of a hard core (the unchanging definitive part of a theory), a protective belt (adaptable theories designed to protect the hard core), and a negative and positive heuristic.  The negative heuristic mandates that the protective belt be sacrificed to preserve the hard core.  The positive heuristic determines how the protective belt should be modified – what should be added to preserve the core without compromising the integrity of the research program.  For example, in order to preserve the core of the Ptolemaic conception of planetary motion in the wake of contradictory evidence, the equant was introduced and further epicycles were added.  While this allowed the research program to survive, it also introduced what Lakotos calls ad hoc changes to the theory, which are auxiliary hypotheses introduced to correct for contradictory evidence but which fail to increase a theory’s predictive power.  Ideally, an auxiliary hypothesis will be substantiated by the introduction of a previously unknown ‘novel fact’ which independently supports the new hypothesis.  Thus initial conditions may be changed to account for the strange orbit of Mercury, but this new hypotheses merely accounts for an anomaly, it is not substantiated through correct prediction of other phenomena.

Lakatos says that a research program is progressing when it consistently adds empirical content (new hypotheses) and intermittently has empirical progression (new, substantiated hypotheses), and a research program is degenerative when it fails to produce both of the above.  A new research program may supersede an old research program if it is progressive while the other is degenerative.  Here Lakatos has married Popper’s conception of falsifiability (as the more complex rejection and refutation) with Kuhn’s general conception of paradigms.

I particularly like Lakatos’ differentiation between rejection and refutation. Having worked in a few research settings, it’s clear to me that being forced to reject any theory that has imperfect data would be mad.  As all models contain assumptions it should be implicit that not all data will fit and that the best model, rather than the right model, should be sought.  Where in vitro studies and computational studies are concerned, the best model is often one that is purposefully incorrect in that it is known to be a highly simplified representation of the true environment which allows for faster and cheaper data acquisition.  I appreciate that Lakotos’ idea of research programs allows for a more applicable philosophic interpretation of science than either Popper or Kuhn’s ideas do.


While I’d like to give my own complex response to each of these analyses of science, I strangely enough feel less comfortable now with the idea of ‘science’ as a concrete thing than I used to.  While I’ve started to question my basic notions of science as a tangible, mathematical field of purely rational evaluations of observable qualities, I don’t yet have a good representation of my current feelings on the matter.  Intuitively in my mind, subjects like quantum physics and biomedicine are scientific in nature, and subjects like intelligent design and psychoanalysis are not, but what differentiates these disciples? Is it that scientific theories are adaptable and aim to best fit the evidence while non-scientific theories instead twist evidence to avoid changes in doctrine? Is it that scientific theories are testable and non-scientific theories are not, is it that scientific theories make predictions and non-scientific theories are ad hoc?  It seems wrong to differentiate the scientific from the non-scientific based on gut feelings, but it also feels wrong to overcomplicate the definition of science to the point of inapplicability.  So for now I’ll leave the question open.






P.S. I finally found the Michigan alumni group in London and I got a chance to get together with some fans at a bar this weekend to watch the Michigan – MSU football game, which was a ton of fun (especially considering we won)! Go Blue!!  🙂




Popper, K. (1963) ‘Science: Conjectures and Refutations’, in M. Curd et al. (eds.) Philosophy of Science: The Central Issues, second edition, New York: W.W. Norton & Company, 2012.

Kuhn, T. S. (1962) ‘The Nature and Necessity of Scientific Revolutions’, in M. Curd et al. (eds.) Philosophy of Science: The Central Issues, second edition, New York: W.W. Norton & Company, 2012.

Lakatos, I. (1968) ‘Criticism and the Methodology of Scientific Research Programmes’, Proceedings of the Aristotelian Society, vol. 69: 149–186. (Read Sections 3-4, pp. 167-186).

Introduction to Philosophy of Science


Hello everyone!

I wrote this two weeks ago and for some reason saved it as a draft instead of posting it, so here it is now:


To help get my own thoughts in order and to jump into this whole “philosophy” thing, I’d like to define a few definitions from my early classes.


Deduction vs. Induction

Deduction is an argument whose conclusion is guaranteed by it’s premises (assuming the premises are true).  An example: It is windy in London.  If it is windy, John is wearing a windbreaker.  Therefore John is wearing a windbreaker.

Induction, in turn, is the act of basing predictions on past observations.  For example: Every time I’ve come near fire, my skin has been burned.  Therefore all fire burns skin.

Whereas deductive reasoning uses evidence to guarantee the truth of the conclusion, induction merely offers support of the conclusion.  I have no way of knowing that every fire is hot without testing it every time, but the notion that all fire is hot is supported by my past observations.

David Hume offers a famous objection to the validity of induction.  First, he refutes the “uniformity of nature” (the idea that nature follows consistent rules that allow the future to be predicted by the past).  As an example of this non-uniformity, Hume offers the example of a chicken that has been raised by a farmer.  No matter that for every day of the chicken’s life it has received food from the farmer, one day the farmer will instead wring the chicken’s neck for his own dinner.  In this instance the chicken inductively relying on the farmer for daily food will one day lead to the chicken’s death.  Nevertheless, Hume admits that some uniformity exists in nature, allowing for often true predictions to come out of inductive reasoning.  His problem with induction comes from its justification.  He says that induction is justified because it often works – often, one can observe, the past can be effectively used to predict the future.  This is in itself a form of induction, and to base the validity of induction on inductive reasoning is a cyclical (and therefore invalid!) argument.

Karl Popper has a response to this, writing that it is wrong to discuss the validity of induction because predictions (or theories) are not in fact based on induction.  Rather, theories are put forth based on preliminary observation, but rather than being “confirmed” by subsequent observations, theories can only be falsified or corroborated – they can be shot down or survive for further trials.  Therefore it cannot be stated that “all fire is hot” is a true statement; instead, one could say that no fire has yet been found that is not hot.

Another example, called “The New Rule of Induction” offers an example of a failure in inductive reasoning:

The new riddle of induction introduces the language L’, identical to English except that blue and green are replaced with “grue” and “bleen”.

An object is “grue” if it is green before time t or blue after time t.

An object is “bleen” if it is blue before time t or green after time t.

Given a series of observations (before time t) that all emeralds found are green (or “grue” in L’), an English-speaking agent would induce that subsequent emeralds will be green.  Similarly an L’-speaking agent would induce that subsequent emeralds will be “grue”, which, after time t means that the agent expects emeralds to appear blue (as after time t “grue” is a blue object).  This leads to translational contradiction, as the two agents, using the same inductive logic, came to two different conclusions about the appearance of future emeralds.

I’m not fully convinced by the new riddle of induction.  I think that, if language L’ existed, agents would not only note the physical appearance of the emeralds in observations, but would also note the time of the observation.  As very few objects change color cyclically over time (the sky being a notable exception), L’-speaking agents would, in my opinion, not be so stupid as to expect the color of stones to change with time just because their language is time-sensitive.  Inductive reasoning would tell them that most “grue” objects which are green before time t (lily pads, grass, etc ..) become “bleen” after time t because they remain green.  Therefore if they observed a series of green emeralds before time t they would expect to see “bleen” emeralds after time t because this how green objects have in the past behaved.


Definition of Science

Another interesting concept I’ve been introduced to is Karl Popper’s definition of science.  While most people would instinctively define science as theories methodically based on observation and tested under constrained conditions, Popper offers a different definition.  He says that science is defined by “risky” predictions – the riskier, the more scientific. Scientific theories must be testable, refutable, and interesting.  A “safe” prediction (you will get sick in the next year) may be testable, but is less scientific than a more specific prediction (you will have a cough starting from November 4th and lasting 11 days).  This of course allows many seemingly non-scientific theories to call themselves science.  A medium can claim to have scientific predictions about the future of your love-life and offer incredibly specific visions, but this does not make them “good” science.  Such theories are generally either easily falsified or are so vague as to make them not very scientific to begin with.



I’ve also included some pictures I’ve taken of London and from my trip to the Cliffs of Dover last weekend.








Information taken from:

Goodman, N. ([1954] 1983) Fact, Fact, and Forecast, 4th ed., Cambridge(MA): Harvard UP, Ch. 3.

Norton, J. (2005) ‘A Little Survey on Induction’, in P. Achinstein (ed.), Scientific Evidence: Philosophical Theories and Applications, Baltimore: John Hopkins University Press, pp. 9–34.

Popper, K. (1963) ‘Science: Conjectures and Refutations’, in M. Curd et al. (eds.) Philosophy of Science: The Central Issues, second edition, New York: W.W. Norton & Company, 2012.