THE PHILOSOPHY OF SCIENCE is to many a collection of old thoughts in dusty books and journals, logical meanderings and debates about semantic definitions. It would appear that to the majority of philosophers of science (those left) who have not abandoned all hope and left for stoicism that schools of thought based on induction are reasonable, or, at least the best we can do.
The problem began when humankind began wondering: are we rational beings? Do we reason differently from animals, and, if so, what are the fundamental aspects of human reasoning that would allow us to call ourselves rationale beings? Popper clearly disallowed even the existence of induction, ending the need for the question on whether induction is rationale, ending, it would seem, the problem best posed by David Hume:
- We reason, and must reason, inductively;
- Inductive reasoning is invalid;
- To reason in a logically invalid way is irrational.
Therefore, we must be irrational.
Hume’s ideas drive us to compare inductive reasoning and deductive reasoning, and Popper’s hypothetico-deductivism and critical rationalism inverts the problems of irrational generalization into a framework of logic that, like pure induction, makes predictions, but, unlike induction, reserves knowledge as that which has survive repeated attempts at falsification. Popper’s solution to the irrationality of induction was to conclude that induction does not exist, which led his critics to search his hypothetico-deductivism for evidence of induction smuggled in somehow.
Some thought Popper evaded Hume. Popper said people who think that do not truly understand his formal calculus of logic. Having read most of what Popper wrote, and considering the proposed alternatives, all of which propose to return science to inductive reasoning, I think it is worth reviewing why those who say that philosophers of science have “moved on” from Popper may be correct, but they are wrong in thinking that Science can benefit in any way by adopting any inductive inference framework in lieu of hypothetico-deductivism.
First, let’s look at Popper.
Given Background Knowledge B, generate Hypothesis H. Think of the most critical test (T) that could falsify H. Conduct T to gather Evidence (E). Interpret H in light of E given B. Update H, and/or B, as necessary.
Now, let’s look at “Inference to the Best Explanation”, or IBE, which some take as an inductive alternative to Popperian Hypothetic-Deductive Reasoning.
Observe phenomenon P. Think of all of the hypotheses H1, H2, H3… Hn that might explain P. Consider lines of evidence that can rule out various Hi’s. Any Hi’s that remain are more probably true because they “best explain” P.
IBE is considered to use induction. But does it? It is really different from H-D? Let’s see:
The Phenomenon P requires and explanation. “The Butler Did It” is one explanation, but it is also an hypothesis H. To test hypothesis H, we try to find reasons (evidence) that can rule it out. How do we find evidence? We conduct a test of some kind, which leads to more observation…. did the Butler have a motive, opportunity and means? Three tests (T1, T2, T3) just one H, but still tests. If the Butler had motive, but not opportunity nor means (say, he was in Hawaii) – well, we can consider the hypothesis “The Butler Did It” falsified. We can estimate the probability of being true as very small.
- In the IBE, the focus has always been on the probability that the most likely and most simple hypothesis explains the data. This makes sense because we all have our own experience in understanding following explanations. However, it does not mean that induction is employed. In fact, a series of T’s to test a series of H to provide a series of E, all interpreted in the context of accepted B is still Popperian logic.
For Popper, induction simply did not even exist. And I agree. What he meant was that valid knowledge claims made from rational inference never involve induction. We call use induction all the time; we generalize, for example, from our experience that we cannot walk on water – any water. But if you’ve never seen a frozen pond or a lake, well that’s a different story. That’s the risk of induction – it can always be wrong up to the point until we have collected, observed, measured, witnessed every last instance and proven the generalization correct. At the very moment when we collect the last observation… induction disappears, because we have the entire set, and we then can safely make our deductive inference, whatever that may be.
The departures from Popper with dalliances with induction have untethered the philosophy of science, and science itself, to a period of irrationality that has arguably facilitated a no-holds barred wrestling match with, and for, very different versions of reality in various domains of inquiry. Partly creating and filling the void first was Kuhn, whose revolutions in science depended heavily on consensus to define knowledge. Of course now, that looks a bit like a popularity contest, and rational thought tells us that so many pieces of information held in esteem as correct by so many (scientific or medical) minds does not actually measure whether those bits of information are, in fact, true; surely there are more rational ways to understand what science is, and how science works than to allow the loudest (or best funded) 800-pound gorilla in the room to carry the day.
Kuhn’s popularity contest encouraged the development of the Skeptics movement, in which public ridicule became an enforcer of reality. Individuals who held notions, ideas, questions, theories or even hypotheses that had been “debunked” were once and forevermore, in the minds of the Skeptics at least, PROVEN wrong, which requires a very strong, positivist induction about absolute truth, which, for me and for many critical rationalists, is unscientific. Science always leaves the door open that what may be a prevailing view today may be found to be incorrect, to a degree or in whole, tomorrow, and while Skeptics say this, in the application of the intentionally pejorative label “debunked”, the Skeptic is playing with inductive fire, because their determination is meant to be definitive, a positive knowledge claim, a universal truth. In reality, of course, they can be wrong on their view that a particular idea has been “debunked”.
Others, notably Musgrave (2004), have analyzed whether, and how, Popper may have slipped up and allowed induction into his logic of science. Musgrave cleverly put the smuggling of induction back on those who accused Popper of using induction and comes very close I think to giving the Popperian framework more than a fighting chance to be taken seriously. However, in a world in which prevailing views are considered valuable, and in which billions of dollars of capital depend on the public – and the scientific and medical community – to accept a particular version of reality, there is little incentive to non-purists to accept or even anticipate a revival of Popperian philosophy.
We must admit, though, that Popper’s main goal was not prescriptive, but was, rather, a description of what science is, and how it is conducted. Scientists even today rely on hypotheses; to dispense with the hypothesis is to cut out the scientist. Companies with products could dispense with the entire charade of Science-like activities and merely publish knowledge claims that go beyond the available evidence – the best headlines money can buy. No, the hypothesis is not dead. Tests of hypotheses – and the attempt to falsify them (in the null form) are ubiquitous among published scientific studies. There are some who lament the over-reliance on the p-value, and its misinterpretation, but certainly the probability that a null hypothesis might be found to be incorrect incorrectly by chance is still relevant. The social aspects of the science as an enterprise – communicating results to peers, interpreting results – can either be seen as an opportunity to attempt to sway prevailing views, or as a responsible reality check against subjective or parochial restrictions. The opportunist, or hack, would prefer the former; the purist the latter.
Popper should be celebrated as a master of the caveats, with expressions that are sometimes missed such as “for the time being”, and “as far as one can tell” and even his concept of “Conjectural Knowledge – but he also bedeviled his readers with inconsistent claims that words (semantics) do not matter, that new terminology should be tolerated, and that his critics did not fully understand him. All of these rallied thinkers to examine for themselves logic and thought processes in a new light, to question the definition of the word ‘belief’ (which Popper famously claimed to not believe in). So, what was Popper up to? Why did he claim that subjective belief played no role – zero – in what he understood as Science?
Popper sought to solve the problem of Demarcation – that is, having a category system in which activities that actually qualified as science could be placed into a bin labeled “Science”, and that other activities could be placed into another bin labeled “Not Science”. It was critical for Popper that Science be rational, that is, as a process, science did not dissolve into what is called the infinite regression, in which peeling back layers of knowledge reveals that each layer is based fundamentally on previous unfounded (irrational) generalizations because they, themselves, were based on inductive reasoning.
A classic description of inductive as generalization used by Popper was the knowledge claim that “All Swans Are White”. The set-up is that if a person has only ever seen White Swans, their experience tells them only that white swans exist. A claim that “All Swans Are White” is inductive inference that goes beyond their experience (beyond the evidence available to them). The claim is clearly not supported by past experience alone, because it involves Swans not yet seen.
For Popper, science progresses beyond induction if one makes a conjecture that is then intentionally subject to refutation by some critical test. A conjecture is an operational claim, not meant to be applied as workable model of reality, that, when subjected to a critical test, is either falsified, or it survives the test. The survival of the test does not confirm the claim; that for Popper was too strong a word; instead, the surviving hypothesis is corroborated, and it acquires a feature he called verisimilitude (meaning closeness or approaching Truth). In Popperian Science, truth is never claimed, beliefs play no role, and induction, if it exist, only exists for a brief, transitive operational form – think tongue-in-cheek – not for the purpose of a knowledge claim, but, instead, in the form of a working hypothesis – something far short of a knowledge claim – that can, in principle, be cut down by evidence from a test, interpreted in light of background knowledge.
The reason why induction does not exist is clear from the Swan example: In hypothetico-deductivism, the claim that all swans are white can be falsified by finding just one black swan. In induction, one can attempt to confirm the claim by pointing to all the white swans one has seen. Finding more and more white swans, however, never confirms the claim, because there may be, hidden somewhere, a lone, black swan. The only solution to truly test the claim is to collect all swans in the world and check their color. When the last swan is collected, at that very point, the inferential exercise flips from inductive to deductive. Induction, it would seem, does indeed not exist (as a form of rational inquiry).
This is why induction does not exist. Generalizations themselves are, by the nature of their existence, eternally subject to being falsified by new observations.
Why This Matters
In the marketplace of ideas, the ideas of science-based or evidence-based medicine carry a lot of gravitas – they appeal to the desire for comfort of knowledge that medical procedures, drugs, and biologics used on patients are being used rationally, that is with a basis found on objective Science. The public, however, has first-hand experience with an ever-shifting understanding of what Science is telling them. Butter is unhealthy. Now, eat butter. Cholesterol is bad. Now it’s not so bad after all. There is a fixed blood-brain-barrier. Oops, no, there is not.
Let’s examine a prevailing view:
“Vaccines do not cause autism.”
Now I’ll inform you that “Wait, not all vaccines have been tested for association with autism”.
That should tell any logician that holding on to “VaccineS do not cause autism” requires use of arbitrary belief, placing it, if you’re Popperian, outside the realm of Science.
And if I inform you that “Association studies do not test hypotheses of causality”, and that “All studies on vaccines that have been examined have been association studies, not randomized blinded clinical trials”, then, if you’re Popperian, you see that the most critical test T has not been conducted.
And then consider that “Some studies do exist that have found higher risk of autism following vaccination” and that AAP and CDC’s lists fail to include such studies, if you’re Popperian, then you might say that AAP and CDC are not constructing a complete and unbiased Background Knowledge, or that they are ignoring the results of Tests which do not fit their arbitrary belief.
And consider “Scientists have for some time suspected that a genetically at-risk subset of individuals exist that have a higher risk of autism from vaccines than others”.
If you’re Popperian, then you might say “Fine, that’s a well-posed hypothesis. Let’s do the Science necessary to Test and attempt to falsify the hypothesis.”
If any of the last four statements are correct, no rational being could possibly conclude “All vaccines do not cause autism”.
But that is what the Kuhnian juggernaut of the CDC has the media telling you. And what likely 99% of all medical professionals will tell you, all the while stating that there have been dozens, hundreds, or maybe thousands of studies that show that vaccines do not cause autism.
They don’t know how many studies. They’ve not read the studies.
Either all of these individuals are misled, are misleading, or have some justification.
At the root of the use of unwarranted generalizations, and willful ignorance of evidence that falsifies those generalizations is a fear – a fear that the public will stop accepting vaccines – if they are informed of the realities of vaccine risk. Fear is irrationality – and the inaction by those who decided to stick our collective heads in the sand for over 15 years tells us that they are… not Scientists.
HHS’s Unfulfilled Duties
It is not exactly clear who decided that HHS would not fulfill the Congressional mandates made in 1986 to make vaccines safer, and to find ways to identify those at highest risk. But HHS never even attempted those tasks. Instead, they undertook (via the CDC and contractees) science-like activities designed to manipulate and control public perception. From 1986 to the present, CDC pushed out studies that found, repeated, no association between vaccination (mostly with MMR) and autism. Not all vaccines have been tested, though, and some studies conducted by others have, in fact found association. Positive results are dismissed, however, because association falls shy of testing causality. And that is precisely where we stand today, and have stood, since the last Institute of Medicine report that made the inductive leap from “vaccines studied” to “vaccines”.
But HHS has a special responsibility in this debacle. Congress mandated them in 1986 to make vaccines safer, and to find those most at risk from serious adverse events. ICANN and Robert F. Kennedy, Jr. received an admission in court from HHS that they have never filed the required biennial report required of them by the 1986 National Childhood Vaccine Injury Act. Instead, they labeled serious adverse events “moderate” on Vaccine Information Sheets. They censored US citizens who participate in their open comment period when they considered weakening the MMR Vaccine Information Sheets because they might scare people away from using the MMR Vaccine. They have conducted cooked studies “to show” that vaccines do not cause autism – all association studies, most very, very poorly designed, off-target, underpowered and manipulated to provide evidence to support the claim
“Vaccines Do Not Cause Autism”.
Science does not put forward evidence to support a hypothesis, it tries to refute it. The genetic x environment hypothesis has never been tested.
It’s time for a return to Science.
Let’s move forward with studies that are aimed at designing models that are shown empirically to detect with high accuracy those children or families who are highest risk of developing neurodevelopmental disorders from neuroimmunotoxins and genetics, and let’s make sure those predictive models are based on profoundly sound and logical hypothetico-deductive inferences.
Postscript: As for Hume, the answer is: Thanks, but it’s a trick question. Humankind can be rational at some times (deductive), and irrational at other times (inductive). Induction is where the risk lies, but knowledge, in the end, always comes from deduction.
Musgrave, A. 2004. How Popper [Might Have] Solved the Problem of Induction – Jstor https://www.jstor.org/stable/3751824?seq=1#page_scan_tab_contents