top of page
  • Writer's pictureJake Browning

Falsification, Paradigm Selection, and Integrated Information Theory

A strange response to integrated information theory (IIT) being labeled a pseudoscience is the claim that "scientists only know Karl Popper's work but fail to remember Thomas Kuhn". The suggestion seems to be that Popper's view that a theory should be "falsifiable" is bad, outdated philosophy of science and, as such, the unfalsifiability of IIT shouldn't be held against it.


This is all a strange interpretation of philosophy of science, in part brought on by ignoring the issues motivating these philosophers. It is helpful to remember the context behind their views--as well as the examples they discuss. Although they are both providing a philosophy of science and disagreeing in part, I think writing off Popper on account of Kuhn is misguided. Kuhn is, as many have pointed out, deeply indebted to positivist philosophy of science, so he's more of a sophistication than a rejection.


It is helpful to remember Popper is, like most positivists, emerging out of the wild, speculative period of the late-19th and early-20th centuries. This was the time of Freud, parapsychology, creative evolution (and whatever Bergson's theory of memory is), strong emergence, parallelism, and a host of other theories that required rather extreme metaphysical posits. But, out of all of these, none worried Popper like the "science of history," the Marxist historical materialism(s) that were used in his day to justify violence against citizens. His critique of unfalsifiable theories connects to his worries about the presumed infallibility of Marxist leaders who often claimed to know the direction of history and punish those on the wrong side of it.


The appeal for falsifiable hypotheses wasn't just a call for theories to be testable; it was also a call for separating the theory from the "scientist"--or, in the Marxist case, the infallible leader. By making theories stand on their own and ensuring they are subject to evidence, it makes clear that disconfirmability is an essential part of scientific practice. The goal was to separate science, as something testable, from the people doing the science, who proved endlessly able to explain away every result or simply regard it as confirming their theory. By focusing on theories as falsifiable, he is arguing scientists need to let the theory speak for itself, rather than have scientists constantly reinterpret every result as either evidence for the theory, or at least not disconfirming.


This doesn't make Popper right, of course. But it does raise an important point about what makes a theory a theory: if a science dependent on the peculiar, ever-shifting interpretations of its practitioners, then no one else can know what the theory means. This reminds us that there are two senses in which a theory can be unfalsifiable: it can make claims that cannot be tested, or the theory is so amorphous and dependent on its practitioners that every result can be explained away.


Kuhn, notably, does not spend time discussing Marxism or Freudianism. His concern is not wild-eyed ideologues engaged in theories that can only be expressed post facto, allowing endless reinterpretations of findings as supporting an ever-shifting theory. But he does focus on scientists as human agents, rather than on theories. But Kuhn's historical investigations point to unquestionably scientific communities that firmly sought empirical evidence, whether confirming or not, focusing on theory-change in physics and chemistry primarily. At least in his major works, he takes for granted that he is discussing respected, and respectable, scientists. The question is how they operate--how scientists do their work--rather than on questions about abstract theories. He defines his paradigms not around theories--as the kinds of things you'd find in a textbook--but around productive experiments and experimental practices that produce fruitful results which can be extended.


As a result, all the scientists Kuhn writes about believe their theories are falsifiable and independent of their interpretations. There was no one (to my knowledge) suggesting phlogiston and ether were brute facts of reason that were impervious to critique, or that other people are incapable of testing their theories unless they share the right a priori understanding of materialism or the unconscious. Rather, the problem was determining how we move from a well-supported scientific practice that is having trouble to a novel, less supported approach.


This is fundamentally about science as a social practice: what do scientists do when they start running into problems ("anomalies") within their regular practice, when experiments produce unexpected, or at least noisy, results? He argued these didn't falsify a theory, but they did mess up the practice of normal science because they show there are areas where our experiments aren't working right--where something isn't going as expected. If enough problems arose, it triggered a crisis--and encouraged people to try something else--that is, a new theory, like Newton's posit about gravity to account for both linear and curvilinear motion. This suggests that all the anomalies can be better explained through a different approach, and that this also suggests a whole host of new experiments to try out.


But Kuhn never suggests empirical adequacy isn't the standard for any paradigm. The whole idea of an "anomaly" is the idea that experimental evidence determines whether a paradigm is still workable or whether the experimental procedures have run up against their limits and empirical adequacy requires rethinking what is going on. Kuhn gives no support to the idea that unfalsifiable claims are an acceptable part of science. He is simply taking it for granted that everyone he is discussing is operating in good faith and agrees the evidence should determine what is true--and, for that reason, can prove theories wrong.


What does all this have to do with IIT? Popper is relevant, of course, since his criticisms of theories being unfalsifiable apply to IIT, since it often defines phi and consciousness in ways that cannot be challenged or tested. But, more than that, he thought theories need to stand on their own feet and be separable from their practitioners is something we have not seen with IIT. They consistently ignore criticisms of the theory, arguing that people are starting from the wrong foot--that the critic needs to accept the IIT starting point to figure out what is really going on. Effectively, there is no theory to criticize, because the practitioners simply move it to meet whatever criticisms arise (which is why we are on our third iteration already).


But, in a real sense, Kuhn provides less support. Kuhn's central innovation is treating sciences as paradigms, defined less by theoretical commitments and assumptions and more around key experiments and similar practices. Notably, there are none for IIT. It simply does not have a scientific social practice, focused on evidential investigation. It has theoretical commitments in search of evidence or even experiments that might confirm it. But the theoretical commitments are basically incompatible with experimental investigation, resulting, as Daniel Burnston recently noted (before the letter, I'll add), that there are numerous different kinds of IIT right now of varying strengths, many of which conflict on what is worth maintaining in the theory.


Reinterpreting Kuhn, we might treat IIT as a "philosophical" paradigm, one centered around solving a particular philosophical problem--the hard problem of consciousness. The varying interpretations of the theory, then, might be seen as conflicting hypotheses within a paradigm of people with a broadly shared starting point. Except this is Burnston's point: the starting point is itself what is in question. At which point, it is clear IIT is already in crisis, torn apart by too many anomalies to hold intact as is. That might be taken as progress: we might, with Lakatos, note the attempts to reinterpret auxiliary hypotheses or adjust some core commitments to save the theory and better address the evidence. It might happen, then, that one of these new IIT theories will prove to even be accurate. There are good reasons for doubt, though; we've already seen a bunch of versions of IIT so far, suggesting there is a deeper problem.


I think it is right to conclude from this that, on neither Popper nor Kuhn's account, is IIT a working science any longer. It is neither well-defined theory independent of its practitioners, or a stable paradigm centered around useful and productive experiments. We might, though, allow that it is a degenerative research program, in Lakatos's sense. We could say, "IIT is a flailing research without any stable, core theory underneath it accepted by all practitioners. It once aimed to be a science, but it is no longer producing anything positive but instead engaged in a desperate, rearguard work to find something defensible worth saying about the idea. It is now in a fight about whether evidence matters and, if it does, what evidence would even look like. Thus, it is a dying science." In which case we could treat it as science--simply science in decline, one unlikely to pass on to the next generation. We might simply assert, "bad science happens. I guess we'll have to wait until those theorists die out or give up."


That isn't really the issue, though, and doesn't explain why many people want IIT treated as different in kind--as pseudoscience. The issue here is that the practitioners of IIT keep promoting it as solid theory, as science not just in good standing, but as the leading theory--perhaps the only theory actually addressing the issues. This incongruity between its front facing persona and its actual status is frustrating if it is subject to evidence and if it is trying to be a genuine science. But it does not seem to be. It seems to be a program that does not feel bound by evidence--that the findings of science are simply not what is at stake. In this sense, it becomes deeply similar to sciences that are unfalsifiable in Popper's second sense: ever shifting theories dependent upon the interpretations of their practitioners.


This is pseudoscience in a common-sense way: something that purports to be a science but is indifferent to evidence. Citing Kuhn or Lakatos helps none at all for this. Calling it a flailing paradigm or degenerative program makes sense if it cares about evidence. But IIT is fundamentally about manipulating social opinion and shifting people's metaphysical assumptions about consciousness. That might be a worthy endeavor, in much the way a proselytizing religion might be, but it is different from anything that actually wants to be accepted on the merits of evidence.


Recent Posts

See All

Why isn't Multimodality Making Language Models Smarter?

Philosophy has something called the "symbol grounding" problem. The basic question is whether the words and sentences of language need to be "grounded" in the world through some sort of sensory relati

AI Doom and the Insurance Argument

In Cixin Liu's The Three-Body Problem, it is discovered that an alien civilization is on its way to Earth, intent on conquering the planet in order to secure its species' survival. Suppose such a thin

Gemini, Grok, and the Crowded Field of Language Models

Tourists on a guided program in Europe often come up with a simple complaint about the area, summed up as ABC: Another Bloody Church. Although each is the pride and centerpiece of their city, they oft

bottom of page