On September 16th, an open letter signed by a group of 124 researchers labeled the integrated information theory (IIT) of consciousness pseudoscience. As Nature reported, this sparked an uproar in consciousness science. Only time will tell whether all the attention will hurt or rather help IIT as prominent neuroscientists (such as Anil Seth) and philosophers (such as Philip Goff) run to its defense. During my post-doctoral research fellowship, I was supervised by a leading IIT proponent, Christof Koch. I have devoted much of my research to constructively critiquing IIT, illuminating some of its well-known and rarely discussed potential pitfalls and possible solutions. Given my research background and awareness of IIT’s strengths and weaknesses, I read the open letter with great interest in the reasons it presents for labeling IIT pseudoscience. Before discussing the rationale offered in the letter, some historical and ideological context will be helpful.
Consciousness Science
As Bernard Baars points out in the introduction of Essential Sources in the Scientific Study of Consciousness, the scientific study of consciousness was taboo during much of the previous century. Excluding consciousness from science, however, meant that science could not study the phenomenon we’re most directly aware of, at least when we’re conscious. Additionally, if the origin of conscious beings is to be given a fully scientific explanation, it seems that a scientific explanation of consciousness is necessary. Hence in the last three decades, the scientific study of consciousness and subfields such as the neurobiology of consciousness have become well established. Yet, at present, there is no paradigmatic scientific theory of consciousness, and for better or worse, there’s no shortage of contenders.
In the late 1980s, Baars was one of the first originators of a neurobiological theory of consciousness, which is now called the global neuronal workspace hypothesis (GNW). This field has since developed; Giulio Tononi first presented IIT in an article published in 2004, and IIT 4.0 was recently published. Although IIT is one of the most prominent theories of consciousness, it has attracted criticism from neuroscientists who find its axioms about consciousness a peculiar place to start neuroscientific research. The theory is based on axioms about the nature of consciousness that are more philosophical than empirical, such as the idea that consciousness exists and has informative structured phenomenal distinctions that are nevertheless integrated or unified. Based on the axioms about consciousness, which are associated with integrated information, postulates about the nature of the physical substrate of consciousness are inferred. Overall, IIT’s postulates suggest that the physical substrate of consciousness will exhibit a maximally irreducible cause-effect structure that is the integrated information consciousness is associated with. IIT is thought to yield the possibility of measuring consciousness via measuring the maximal irreducible causal structure manifested by a physical system. This is a measurement of what is called Phi, which is symbolized by Φ, and it can be applied to any physical system, according to IIT.
Although IIT invites criticism from some neuroscientists, it is important to understand why it’s attractive to others. At the end of the 20th century, Francis Crick and Christof Koch instigated the search for the neural correlates of consciousness (for brevity NCC), which is foundational to the neurobiology of consciousness. At this year’s ASSC meeting, NCC research attracted significant attention due to David Chalmers winning a bet with Koch, who wagered that by June 2023 the neural correlate of being conscious—technically called the full NCC—would be discovered. While neural correlates of content-specific conscious states such as seeing a face are easier to identify, the full NCC remains incognito, as its brain location and neuroanatomy remain unverified. As the research continues, IIT advocates look to the theory’s postulates for guidance (and proponents of other theories use the same method but different predictions implied by their preferred theory).
IIT’s postulates guide the neuroscientific research by indicating where in the 86 billion neurons of the brain the neural correlate of being conscious is most likely to be, suggesting it’s in a temporal-parietal-occipital hot zone that has sufficient neural architecture for reciprocal projections capable of manifesting a maximally irreducible cause-effect structure. By contrast, the global neuronal workspace hypothesis predicts consciousness corresponds to a global broadcast of information mediated by a prefrontal-parietal network of long-range cortical neurons. Since the details are daunting, the competing predictions are often casually described as corresponding to the front versus the back of the brain, even though the neuroanatomy is not so clean-cut.
Rationale for Labeling IIT Pseudoscience
The letter labeling IIT pseudoscience was drafted and/or signed by some prominent neuroscientists and philosophers who typically prize rigor and clarity. Yet, despite the seriousness of the charge being made, it contains no definition of science nor pseudoscience. This omission has been widely critiqued and Hakwan Lau, one of the authors, has attempted to remedy it. A statement promising further explanation in an article was added to the top of the letter “following feedback from the community” after the letter’s release, which some consciousness science researchers fear could damage the whole field. Yet, as the letter currently is, the lack of clear definitions that all authors endorse makes it difficult to analyze how the reasons offered in the letter support its conclusion that IIT should be labeled pseudoscience. Nevertheless, it is worthwhile to analyze some of the primary reasons offered in the letter itself (and perhaps my brief analysis here can provide points for the letter authors to consider as they draft their article).
To begin with, the letter offers what we can call an epistemological concern. The letter critiques experiments testing IIT against its “adversary” GNW, in a large-scale adversarial collaboration. The letter states:
The experiments seem very skillfully executed by a large group of trainees across different labs. However, by design the studies only tested some idiosyncratic predictions made by certain theorists, which are not really logically related to the core ideas of IIT3,6,7, as one of the authors himself also acknowledges8.
The authors go on (see the lengthy quote below) to suggest that the tested predictions are shared by other theories and are therefore not unique to IIT. Like others, I have also critiqued the rationale for inferences based on IIT’s tenets (see Owen 2019, pp. 181-183; 2021, pp. 180-182). I think the above critique is perfectly reasonable to raise.
But it is likewise fair to ask: What is the logical relationship between this critique and the conclusion that IIT is pseudoscience? The critique is about the epistemic relationship between experimental predictions and the core ideas of IIT. This is worth analyzing because the relationship may be one of logical necessity, probability, or explanatory adequacy. What can be justifiably inferred on the basis of the predictions being confirmed in the referenced experiments depends on exactly how the tenets of IIT informed the predictions, and whether they necessarily or probably follow, are explained well by IIT’s tenets, or are merely consistent with such tenets. Such relations correspond to different types of arguments. Successful deductive arguments have conclusions that follow from their premises with logical necessity and inductive arguments have conclusions that follow with a high degree of probability. In the case of abductive arguments, warranted conclusions are considered the best explanation of the premises or data, even though there can be other possible explanations that are just as logically consistent with the data.
What’s interesting about the above critique is that the authors seem to be assuming that if the confirmed predictions do not follow with logical necessity from IIT’s core tenets and are consistent with not only IIT but also other theories, that’s a problem. It’s not clear, however, why this is a problem, much less one that qualifies a theory as pseudoscience. It would be a problem if IIT could only argue deductively from its core tenets to the predictions, but IIT theorists can formulate their reasoning abductively, as inferences to a good explanation (see Albantakis et al. 2023, p. 6). Abductive reasoning is common in science, and the conclusion of an abductive argument is a theory that may compete with other theories to explain a data set. Since multiple theories can be logically consistent with the same data set, the question often becomes: which theory best explains the data? Yet, given that multiple theories can be consistent with the data, or make the same predictions, albeit with differing degrees of probability, it is hardly surprising that IIT might inform predictions that are consistent with other theories and positions (cf. Evers, Farisco, & Pennartz, forthcoming). While this can make it more difficult to determine which theory enjoys greater justification, it’s difficult to see how this in any way implies that IIT is pseudoscience, especially without implying that numerous scientific theories are pseudoscience. Exactly how such a serious charge is epistemically related to the proffered critique needs clarification just as much as IIT’s relationship to the tested predictions.
As we move on to what appears closer to the heart of the letter’s rationale for labeling IIT as pseudoscience, it will be illuminating to quote the letter at length:
According to IIT, an inactive grid of connected logic gates that are not performing any useful computation can be conscious—possibly even more so than humans17; organoids created out of petri-dishes, as well as human fetuses at very early stages of development, are likely conscious according to the theory18,19; on some interpretations, even plants may be conscious20. These claims have been widely considered untestable, unscientific, ‘magicalist’, or a ‘departure from science as we know it’15, 21-27. Given its panpsychist commitments, until the theory as a whole—not just some hand-picked auxiliary components trivially shared by many others or already known to be true28-31—is empirically testable, we feel that the pseudoscience label should indeed apply. Regrettably, given the recent events and heightened public interest, it has become especially necessary to rectify this matter. If IIT is either proven or perceived by the public as such, it will not only have a direct impact on clinical practice concerning coma patients32, but also a wide array of ethical issues ranging from current debates on AI sentience13 and its regulation, to stem cell research, animal and organoid testing18, and abortion19. Our consensus is not that IIT and its variants decidedly lack intellectual merit22. But with so much at stake, it is essential to provide a fair and truthful perspective on the status of the theory. As researchers, we have a duty to protect the public from scientific misinformation.
An analysis of the claims concerning public relations can be found elsewhere. The only thing I will say here is that I agree with the authors of the letter with respect to their desire for IIT to be accurately represented, especially to the public. Yet, such a desire should also motivate an analysis of the reasons offered for labeling IIT as pseudoscience. Because accurate representation of the theory is important, it should be acknowledged that there are more options available beyond the labels of either ‘proven’ or ‘pseudoscience.’ There are numerous theories of consciousness that are not proven, but that doesn’t mean the only option is to label them pseudoscience. An obvious option is to simply call them ‘unproven.’ Given that this option is on the table, calling the theory pseudoscience requires sufficient justification.
As a philosopher, I was struck by how heavily the rationale offered in the lengthy quote above relies on philosophical disagreement. At the beginning, the authors critique IIT for predicting that an inactive grid of connected logic gates could be conscious even if it performs no useful computations. Assuming consciousness requires a bearer—i.e., the subject who is conscious (see Guta 2019, p. 132)—which logic gates lack, the idea that such could be conscious strikes me as unreasonable. Notice that the letter writers likewise disagree with IIT for a philosophical reason. They think useful computations are necessary for consciousness. This assumes a particular view in philosophy of mind—a computational functionalism—that IIT fundamentally disagrees with see (Koch 2019). It’s great that the authors of the letter disagree with IIT; the more ideological diversity the better for fruitful debate. But, once again, the reader of the letter is forced to wonder: how does IIT’s divergence from a computational view of the mind render it pseudoscience?
IIT is also critiqued for embracing panpsychism. I, too, disagree with the panpsychism which is often and unfortunately associated with IIT (see Owen 2019, pp. 178-183). Nevertheless, to infer that IIT is pseudoscience based on its quirky position in philosophy of mind comes across as more of a smearing tactic resembling contemporary American politics than a justifiable reason for attaching the ‘pseudoscience’ label. By all means, the authors of the letter could provide philosophical arguments against panpsychism, which would be far more respectable than trying to undermine IIT by labeling tactics. And there are solid objections to panpsychism one could employ toward such an end (see Rickabaugh & Moreland 2023, chapter 6).
Let’s move on to the bioethical rationale appealed to in the above quote. The authors cite the implications of IIT with respect to bioethical issues concerning abortion and organoids, as well as AI. Again, to sound like a broken record, I agree with the letter writers—IIT certainly has implications for such topics. But this is likewise true of other theories of consciousness, which a leading luminary behind the letter—Hakwan Lau—is well aware of. He has applied his preferred theory, GNW, in a published exchange with other neuroscientists such as Koch and Victor Lamme pertaining to machine consciousness and ethical implications. Regarding the topic of abortion, the authors of the letter appear concerned that IIT suggests fetuses could be conscious at early stages of development. This would favor the pro-life position, which prominent signers of the letter such as Lau and Churchland might disagree with. Yet, should a theory be branded ‘pseudoscience’ because it has implications that appear to support the pro-life position in the abortion debate? If so, should scientific studies demonstrating the self-organization of embryos also be labeled pseudoscience if they have ethical implications for the abortion debate that appear to favor the pro-life position? Whatever one’s position is on this contentious bioethical topic, it probably shouldn’t be the standard for what counts as science or pseudoscience. However, if it should, then the response to the letter from Victor Lamme, the originator of the recurrent processing theory, should be quite sobering for advocates of multiple prominent theories.
The letter also cites potential implications for organoids. The fact that neurobiological theories of consciousness could have such implications is precisely why I have teamed up with colleagues to theoretically apply such theories to human cerebral organoids. Personally, I hope human cerebral organoids cannot possibly be conscious. I hope we can responsibly and justly continue to use them to better understand the brain, and to make neurological progress. However, it’s not up to me, or anyone else, to decide whether they can be or cannot be conscious. It is rather something for us to discover, and we should utilize the best tools we have to make such discoveries. In other words, we should consider how prominent theories of consciousness can be useful, regardless of whether they predict conclusions we agree with or disagree with.
Once again, it’s difficult to see why the authors of the letter think that IIT’s potential implications regarding organoids are rationale for labeling the theory pseudoscience. Such a label is more likely to stifle serious responsible analysis of IIT and its implications than promote such analysis. If IIT or any worthwhile theory has potential implications for ethical actions and policies, why and how will labeling it pseudoscience be helpful? Wouldn’t it be better to do all we can to examine what exactly the theory implies and to test whether or not the theory is actually true? If such information could have real-life ethical application, that seems more like a reason to consider such information and to avoid doing anything that might silence it.
I worry that labeling a theory as pseudoscience is more likely to lead to it being discarded simply because of a label, rather than because it was analyzed and tested. The result would be that its potential ethical implications would more likely be ignored than adequately interrogated. That’s the opposite of ethical and scientific responsibility. Those of us who receive paychecks from publicly funded institutions to do research owe it to taxpayers to investigate plausible theories with ethical implications even if those implications run contrary to our own philosophical and political opinions.
Is IIT true? I doubt it. Is it pseudoscience? Perhaps, but one would have wished for better reasons supporting such a claim. IIT should certainly be critiqued and, most likely, in due course disproved and therefore set aside. However, if it is discarded because of receiving a label, that would resemble the sad state of American political tactics more than science. To give an able critic of IIT, Tim Bayne, the last word:
The charge of pseudoscience is not only inaccurate, it is also pernicious. In effect, it’s an attempt to “deplatform” or silence integrated information theory – to deny it deserves serious attention. That’s not only unfair to integrated information theory and the scientific community at large, it also manifests a fundamental lack of faith in science. If the theory is indeed bankrupt, then the ordinary mechanisms of science will demonstrate as much.
Matthew Owen
Matthew Owen (PhD, University of Birmingham) is a faculty member in the philosophy department at Yakima Valley College in Washington State. He is also an affiliate faculty member at the Center for Consciousness Science, University of Michigan. Matthew’s latest book is Measuring the Immeasurable Mind: Where Contemporary Neuroscience Meets the Aristotelian Tradition.
