Must an hypothesis be falsifiable to be scientific?

  • Clarifying the question

    According to Popper, the idea of falsifiability corresponds to the idea that a statement “must be capable of conflicting with possible, or conceivable observations.”1 It is easy to demonstrate that an hypothesis being falsifiable is not a criterion for its being true, rational, or belief-worthy.2 However, some still feel that it is nevertheless a criterion for its being scientific. But is that right?

    1. Karl Popper, Conjectures and refutations. The growth of scientific knowledge, (Basic Books, 1962), 39.
    2. For examples of propositions can be rational to affirm, but nevertheless unfalsifiable, consider these: • It is possible for the world to exist • You are not in the Matrix • Other persons are conscious, rather than p-zombies • 2+2=4 etc.
  • Scientists and Philosophers say “NO”

    It is well known among philosophers of science, to the point of being in intro textbooks, that hypothesis need not be falsifiable to be considered scientific. However, the fact is also recognized often enough by scientists themselves and their organizations:

    • Scientific American: “It should be noted that the idea of falsifiability as the defining characteristic of science originated with philosopher Karl Popper in the 1930s. More recent elaborations on his thinking have expanded the narrowest interpretation of his principle precisely because it would eliminate too many branches of clearly scientific endeavor.” [John Rennie, “15 Answers to Creationist Nonsense.” Scientific American (July 17, 2002).]
    • NewScientist: “Karl Popper famously took on this so-called ‘demarcation problem’in the 1930s, arguing that scientific hypotheses differ from pseudoscientific ones in that they can be falsified through experiment. Today, many scientists continue to see falsification as the hallmark of good science (physicist Leonard Susskind refers to them as the ‘Popperazzi’), despite the fact that philosophers have long since realized that science can't be wrapped up quite so neatly.” [Amanda Gefter, “Tracing the fuzzy boundaries of science.” New Scientist (May 24, 2010)]
“No, after all…
  • Some clearly scientific statements are unfalsifiable

    Consider a few propositions which are clearly unfalsifiable, and yet which nevertheless are scientific:
    • Other universes exist
    • Aliens exist
    • Electrons exist

  • Some clearly scientific hypotheses regularly tolerate falsification

    Some clearly scientific hypotheses regularly tolerate falsification (especailly in the so-called complex sciences, like medicine).1 This is relevant because the these sciences are quintessentially scientific.

    • “Falsifiability is a myth. In the messy real world of observations and experiments, theories often must be modified to accommodate new evidence. Since any theory can be modified with ad hoc assumptions to agree with contradictory observations or experiments, no theory is strictly falsifiable whether it is called science or not. Criteria for judging a modification to a theory unreasonable and therefore rejecting the modification must be added for falsifiability to have any real meaning. Falsifiability is often a double standard. Politically unpopular, unorthodox, or simply new theories are demanded to be strictly falsifiable, a condition they can never meet. Accepted scientific theories are asserted to be falsifiable even though they have been and are frequently modified to explain contradictory data. The need for criteria to reject a modification to a theory as unreasonable is either denied or not mentioned, creating an illusion of certainty. These rejection criteria are not mentioned because they are often matters of fallible personal judgment and opinion. The doctrine of falsifiability offers only a mirage of certainty in distinguishing science from non-science.” [John McGowen, The Myth of Falsifiability (2000).]
    1. The idea here is that one can always make their theory falsifiable by keeping it alive with falsifiable “auxiliary hypotheses” that account for exceptions in ad hoc ways.