The Bat and the Ghost

Official: “Nonsense! No matter what you say, you’ve no proof that you’re a life-form!”

The Puppet Master: “It is impossible to prove such a thing. Especially when neither modern science nor philosophy can explain what life is. (…) I am not an A.I. (…) I am a living thinking entity that was born in the sea of information.”

Project 2501 Scene

Ghost in the Shell (1995),

screenplay by Kazunori Itō,

directed by Mamoru Oshii

This is a story about two seemingly different characters. As in any good story, these characters are driven by deep motivations. The first is the well-known Cartesian ghost described by Gilbert Ryle. This familiar character is the old archetype of dualism— the view according to which fundamental reality splits into, at least, two domains, the mental and the physical. The ghost was conjured by Descartes around the 17th century to explain the nature of thinking things, which, from his perspective, couldn’t arise from the purely mechanical arrangement of the physical world. Descartes was a rationalist; to him, the lack of intelligible or a priori connections between the properties of thought and the physical world justified such a divide. But then a major problem arose: we are also physical beings in whom both aspects of reality appear to interact. Descartes’ solution was to posit a brute connection between mind and body. To many scholars, the main trouble with the Cartesian ghost was the dualist divide. To me, the main issue was demanding intelligible connections between mind and body while accepting their brute connection as a solution. In other words, the ghost isn’t really dualism, but inconsistent rationalism.

Arguably, with the developments of computability theory and its implementation in the cognitive sciences, the nature of thought is no longer a mystery. There are details to be worked out, but it’s plausible to claim that, for all intents and purposes, thinking is computation. So, the Cartesian ghost has been exorcised from our contemporary scientific worldview, hasn’t it? Here is where our second character enters the stage, the bat. This character is the embodiment of what philosophers call “the hard problem of consciousness”. This is a fairly recent issue, but it is at the heart of the mind-body problem. The bat was introduced by Thomas Nagel in the 70’s, and it’s still here to remind us that there’s something it is like to be ourselves and that this subjective character is constitutive of consciousness. While the ghost’s obstacle was based on the early modern limitations to explain thought, the bat’s problem is about consciousness and appears to be the ultimate challenge for scientific explanation. It doesn’t matter whether we’ve figured out the origin of the universe, the molecular basis of life, or the computational nature of thought; many philosophers declare that the subjective character of consciousness cannot be explained using typical physicalist resources without substantial explanatory gaps, or so it seems.

Before going any further, let’s step back and ask: what’s wrong with explanatory gaps? An explanatory gap in some domain poses a problem only if, to make theoretical progress in that domain, we ought to explain all the facts. But we all agree that explanation must stop somewhere. So, why can’t there be brute necessary connections between consciousness and the physical states of our brain? To be clear, many dualists nowadays accept the existence of brute contingent laws that relate consciousness to neurobiological states; what they reject are brute necessary connections between consciousness and those physical facts. So, the key question is: why can’t necessary connections be brute facts in a theory of consciousness? I submit that, as in the ghost’s case, the bat’s problem seems to be rooted in some sort of rationalist metaphysical bias. To put it bluntly, the bat is the ghost. This is the main thesis defended here.

To arrive at my conclusion, I’ll argue for two related radical ideas, one historical and one methodological. As I advanced earlier, the historical statement is that Descartes’ core error wasn’t dualism, but inconsistent rationalism. I’ll provide evidence that this rationalist bias is very much present in contemporary Western philosophy of mind. The methodological implication derived from this is that to make real progress in our reflection about consciousness, we should give up our obsession with the question of physicalism (is consciousness fundamentally physical?) and focus on the question of rationalism instead (are brute necessary connections ruled out from a plausible theory of consciousness?). I end with some suggestions on how to implement this paradigm shift by drawing insights from the current revolution on artificial intelligence (AI).

1. Descartes’ Real Error

This story has been told way too many times, but I’ll add a twist to the traditional plot. We start from the middle— with Descartes having established that the essence of the mind is being a thinking thing and trying to argue for the real distinction between mind and body. He has given us different arguments to that effect; however, one particularly salient is the conceivability argument introduced in the Sixth Meditation.

Descartes’ conceivability argument begins with the rationalist claim that if I know my essence, then I can conceive (or, in his terminology, “clearly and distinctly understand”) my mind existing without my body. Presumably, I can know a priori that my essence is being a thinking thing —the conclusion established by the famous Cogito argument in the Second Meditation. Thus, it follows that I can conceive my mind existing without my body. But if I can conceive such a thing, according to Descartes, then it’s genuinely possible for my mind to exist without my body. And, if that’s a genuine possibility, then they are not the same (“my ability clearly and distinctly to understand one thing without another suffices to make me certain that the one thing is different from the other.¹”). This moment marks the grand conjuration of the ghost. Now, to complete the ritual, Descartes had to say how the ghost is related to the mechanics of the body. Their relation was for him one of the two things that supported our understanding of the nature of the mind. More precisely, in his correspondence with Princess Elisabeth of Bohemia, he wrote: “there are two things about the human soul on which all the knowledge we can have of its nature depends: one of which is that it thinks, and the other is that, being united to the body, it can act on and be acted upon²”. This passage both asserts the mind-body unity and acknowledges their causal interaction.

However, as Elisabeth pointed out in her reply, there is a tension between the real distinction and mental causation. How could something that lacks extension and whose essence is thinking (i.e., the ghost) have causal powers over a body if causation requires physical contact? According to the official doctrine³, the dualism coming from the ghost is the core problem of the Cartesian philosophy of mind.

I disagree with this narrative. While dualism might lead to some problems, it isn’t the problem. Here is where I depart from conventional wisdom.

I claim dualism isn’t the real ghost because Descartes himself provided an answer to the issue. His proposal was utterly simple: take the union and interaction between mind and body as a metaphysical primitive⁴ or fundamental, that is, it cannot be explained in terms of anything else. So, if we have a solution, why don’t people buy it? The answer lies in the solution’s apparent arbitrariness. As a rationalist, Descartes was committed to intelligibility (availability to understanding), a central tenet of early modern philosophy. Arbitrariness conflicts with this commitment. While rationalism encompasses diverse doctrines, one thesis was fundamental to early modern rationalists: the principle of sufficient reason (PSR). Spinoza used it as a core assumption in his system. For instance, PSR serves as a premise in his cosmological argument for God’s existence⁵. Similarly, Leibniz appeals to PSR to derive his famous principle of the identity of indiscernibles⁶. But what makes PSR so appealing? One hallmark of rationalism is the explanatory role of a priori knowledge and intelligibility⁷. PSR extends this constraint to reality as a whole, more precisely, according to PSR, necessarily, for any fact, there is an intelligible explanation of why that fact obtains. In other words, PSR rules out brute facts. The only unexplainable facts that PSR tolerates are those always available to reason—facts that obtain as matters of metaphysical necessity and are knowable a priori⁸ (i.e., without the need for empirical observation). To many rationalists, this is the case of God’s existence, but definitely not the interaction between mind and body. The problem is that Descartes’ primitivist solution takes the union of mind and body to be a brute fact. This violates PSR.

But what if Descartes wasn’t fully committed to rationalism? The difficulty is that in the third Meditation, he wrote: “it is indeed evident by the light of nature that there must be at least as much [reality] in the efficient and total cause as there is in the effect of that same cause. (…) Hence it follows that something cannot come into being out of nothing, and also that what is more perfect (…) cannot come into being from what is less perfect⁹”).

Descartes’ version of PSR (“something cannot come into being out of nothing”) is closely connected to his view that causation itself has an intelligibility requirement (“there must be at least as much [reality] in the efficient and total cause as there is in the effect of that same cause¹⁰”). This is hardcore rationalism, to say the least. The inconsistency is clear: one cannot endorse a rationalist principle like this about causation and then claim that the causal interaction between mind and body is a brute fact. In other words, the historical twist is that the Cartesian ghost was never really dualism; it was bad rationalism. You might wonder why this unorthodox reading matters beyond historical scholarship.

2. Bat Rationalism

The answer is that such a ghost is still with us —though in a chiropteran form— and represents today's main obstacle to making progress on the mind-body problem. Before arguing for that, however, let me say a few words about one scientific development that temporarily expelled the ghost: the theory of computation.

Computability theory, and particularly the development of computing devices in the 20th century owing to the mathematical contributions of Turing, Church, and Von Neumann, marked the pinnacle of rationalism. Modern computers showed us that a mechanical implementation of rationality was possible. Moreover, they showed that it was physically possible. Even in simple physical systems, propositional calculus can be implemented through tiny electric switches controlling binary signals. The secret ingredient is that the causal structure of the system must mirror the logical structure of the program. In other words, you take care of the physics and, voilà, the logic follows. Furthermore, computational power becomes a matter of scale. If you want more and faster operations, just get more—and more sophisticated—switches. This both achieves Leibniz’s dream of a mechanical implementation of rational processes and meets Descartes’ challenge of intelligent attributes arising out of purely physical systems. The insight is that something along these lines applies to the mind-brain such that once the causal structure of the brain is fixed, the rational structure of thought follows.

Suddenly, the ghost disappeared. Thought was physical computation all along, and the human mind just a sophisticated pattern written in our nerves— or better, with our nerves. At least, such was the promise of a doctrine known as ‘functionalism’. One of the merits of functionalism and the computational theory of mind is that it captures the abstract and rational nature of thought without any ghostly substances and only appealing to the materialistic resources already available in our scientific worldview. In other words, it’s a physicalist theory. Physicalism is the worldview according to which all facts are either dependent on fundamental physical facts or are fundamentally physical themselves. As such, any physicalist theory has rich but constrained explanatory resources ranging from fundamental particles to the functional organization of the brain. Of course, there are some worries associated with the computational theory of mind that should be addressed¹¹, but for the most part, when it comes to intelligence and cognition, the functionalist program delivers. Our current AI revolution is living proof.

The main issue is that even if functionalism gets rid of our old Cartesian ghost, the mental domain isn’t exhausted by abstract intelligence. Our mental life is filled with rich conscious experiences marked by qualitative features. Consider sensory or emotional experiences like feeling¹² pain, feeling happiness, tasting red wine, smelling durian, seeing the color palette of a Rothko painting, etc. These conscious states are marked by a certain subjective character. In the 70’s, Thomas Nagel wrote an article that changed everything¹³. His paper made at least two substantial contributions. First, it provided an intuitive way to describe the subjective character associated with those conscious states: that there is something it is like to be in those states for the subject (e.g., hence the iconic question What is it like to be a bat?). Second, it emphasized the explanatory challenge that arises from conscious states, that is, the subjective character isn’t accidentally related to them; it seems constitutive of them. Using the famous example, it’s not just that, given that we lack the bat’s brain, we can’t know what it is like to be a bat. Nagel certainly would agree with that, but the issue isn’t merely epistemic. The issue is rather this: making sense of what it is like to be a bat (“to form a conception” in Nagel’s terms) just is to take the bat’s point of view. In other words, to explain consciousness, the subjective aspect isn’t just a coincidence; isn’t merely necessary, it’s essential to it¹⁴.

Why is this a problem for functionalists¹⁵ (and for physicalists more broadly)? The challenge arises because the correlations that neuroscientists continue finding between our conscious states and the functional states of our brains make no progress on this issue. It wouldn’t even matter if such correlations became causal or obtained as a matter of necessity (or “supervenience” as we philosophers like to say). Not even if we posit that conscious states are identical to brain states (psychophysical identity theory). As long as those brain states don’t help us make intelligible what it is like to experience such and such (“to form a conception” in Nagel’s words), they leave an explanatory gap. If there is such a gap between consciousness and the brain, then they don’t inhabit the same ontological realm. And the problem is, there seems to be such a gap. Thus, although the ghost may have left the machine, the bat takes its place.

The situation gets more critical. Consider that hundreds of scientists around the world (including some Nobel Prize winners) have tried and keep trying to solve this riddle. The stubbornness is partially justified. Many believe that our scientific worldview is incomplete without a physicalist account of human consciousness since this approach has worked well in everything considered providential in the past: the origin of the universe, life, and intelligence. So, there are inductive grounds for optimism. But until we get rid of the bat, completing this task appears impossible. The main objection against any physicalist attempt will always be the same: it leaves an explanatory gap. That’s why philosopher David Chalmers has called this ‘the hard problem of consciousness¹⁶’.

Nevertheless, there is a crucial step that is often overlooked in this piece of reasoning. Let me give you an example from a different domain on a more cosmic scale. Let’s suppose we go back in time to the beginning of everything — to the initial boundary conditions, the primitive physical state of the universe (or the multiverse, if you believe in such a thing). Whatever that turned out to be, let’s call it ‘φ’. It seems legitimate to ask, what explains φ? Ironically, it also seems legitimate to answer: nothing — we have to start somewhere, and φ is just how things turned out to be. However, to many, it seems less plausible to answer the following: nothing — we have to start here, and φ is just how things must be. Why this asymmetry in our φ-judgments?

Prima facie, the issue can be described by distinguishing between contingent brute facts and necessary brute facts. By a ‘contingent fact’ we mean facts that can fail to obtain or could have been otherwise (e.g., it’s contingent that grass is green); by ‘necessary facts’ we mean that they must obtain regardless — things couldn’t have been different (e.g., it’s necessary that green is a color, that 2+2 =4, etc.). Now, given today’s physics standards, it seems acceptable to claim that φ is a brute fact; that is, there is no explanation of why φ obtains. But given today’s metaphysics, the previous claim is acceptable 12 only if we acknowledge that things could have been different. In other words, it is possible that everything started with another state of affairs —say ψ— instead of φ. Now, consider the stronger claim that φ is how things must be with no alternative. Here φ not only obtains without explanation, but also as a matter of necessity. To many, we can’t accept necessary facts like φ because there seems to be no reason why the physical state φ should have the status of a necessary beginning. There is no intelligible connection between the physical fact φ and the “necessary beginning fact”. Therefore, there is an explanatory gap between them. To be clear, the problem isn’t simply the modal status of φ, namely, that it obtains by necessity. Throughout history, many philosophers have been willing to appeal to necessary beings like God. But they did so because, unlike physical state φ, there's an intelligible connection between God's existing and a necessary beginning. On a plausible understanding of the concept of God, it a priori entails a necessary beginning (even an atheist would admit this). Thus, one lesson from our case is that explanatory gaps get closed by intelligibility relations between the concepts at stake.

You may think that perhaps the hidden assumption was that all necessary facts must be intelligible, that is, there are no brute necessities. This gets us very close, but we need some refinement. Many philosophers have convincingly argued that unexplained necessities are acceptable as long as they are like primitive definitions (e.g., axioms) or like essences discovered a posteriori (e.g., water = H2O). It would be useful to put a label on the domain of necessary facts outside of those cases. For lack of a better term, let’s call them ‘a priori facts’. So, the assumption is really this: there aren’t brute (i.e., unintelligible) a priori facts. Now we can see the problem with physical state φ as the necessary beginning of the universe. The necessary beginning of the universe (an alleged a priori fact) isn't made intelligible by physical state φ. In other words, such would be a brute a priori fact¹⁷.

I won’t argue against that assumption. What I want you to notice instead is that it isn’t an innocent claim— it’s a rationalist doctrine akin to a modern defense of the principle of sufficient reason (PSR). It should come as no surprise that Descartes, Leibniz, and Spinoza all appealed to similar principles to defend God’s existence. With that in mind, let me introduce the core claim of this essay: the hidden assumption in the hard problem of consciousness is precisely this form of rationalism. Recall, the problem is the explanatory gap between consciousness and the brain because brain states don’t make intelligible what it is like to experience such and such. But that's only a problem if you assume two things: that facts about consciousness are a priori facts, and that a priori facts can't be brute. So, you get a hard problem only by assuming the rationalist principle that there can’t be brute a priori facts in the domain of consciousness. Call this view ‘phenomenal rationalism’.

Phenomenal rationalism is very present in the classic formulation of the hard problem. So, it’s quite surprising that it’s never made explicit. In his original statement of the explanatory gap, Joseph Levine writes: “The identification of the qualitative side of pain with C-fiber firing (…) leaves the connection between it and what we identify it with completely mysterious. One might say, it makes the way pain feels into merely a brute fact¹⁸”. This is phenomenal rationalism at its clearest.

Similarly, Chalmers’ original formulation of the hard problem when discussing explanations based on the brain’s information processing is this: “Why doesn’t all this information-processing go on in the dark, free of any inner feel? We know that conscious experience does arise when these functions are performed, but the very fact that it arises is the central mystery. There is an explanatory gap (..) between the functions and experience¹⁹”. This has some resemblance to the cosmic question, “why is there something rather than nothing” — or in terms of our previous example, why can’t φ be the necessary beginning fact? This resemblance, I submit, is no coincidence. To say that brain computation can’t ground consciousness without intelligible connections between them is like saying that φ can’t be the necessary beginning fact because there can’t be brute a priori facts. Furthermore, as Chalmers admits, the most celebrated thought experiments against physicalism —like the case of Mary the superscientist²⁰ or the philosophical zombies— are ultimately based on the apparent failure of a priori entailment from brain states to conscious states. ‘Failure of a priori entailment’ is just a fancy way of saying that there aren’t intelligible connections at stake. Again, this is phenomenal rationalism.

In the previous section, we saw that the ghost was rationalism all along. We can see that, though they are centuries apart, the bat carries the same rationalist bias in disguised form. In a slogan: the bat is the ghost. With this perspectival shift, it appears that to make progress on the hard problem, the fundamental question becomes: Is phenomenal rationalism true?

3. The Obsession with Physicalism

Unfortunately, I don’t know the answer. But I have a couple of suggestions to offer. The main one can be implemented as a methodological strategy: prioritize the question of rationalism over the question of physicalism. For all I know, this has not been done. In fact, my impression is that philosophers and neuroscientists remain obsessed with physicalism. Ironically, focusing on the problem of rationalism may end up helping us settle the question of physicalism once and for all.

Entering today’s discussion in the philosophy of mind is a bit like stepping into an argument in an old marriage. It appears that everything that could be said has been said already, but somehow the field seems stuck. To be fair today's trends are much more open-minded about theories once considered ‘too exotic’, like idealism or panpsychism²¹. This shows some progress since these views deserve to be taken seriously, but they remain embedded in the old dialectic as alternatives to physicalism. I think that to make genuine progress, we should give up our obsession with physicalism completely. The reason is straightforward: physicalism is orthogonal to the hard problem of consciousness, and what’s lurking beneath is ultimately rationalism. Remember, the bat is the ghost. To illustrate this, I'll examine two contemporary non-physicalist attempts to solve the consciousness problem as case studies.

First, let’s have a look at a contemporary and influential version of dualism, like the one defended by Chalmers²². Against all odds, this theory is a naturalistic type of dualism in the sense that it claims that a scientific theory of consciousness is possible. The dualistic element lies in the claim that such a theory can’t be based on reductive explanations of any sort. Consciousness is fundamental; it’s not grounded in anything else. So, the puzzle becomes: how do we explain the psychophysical correlations that neuroscience reveals to us? For naturalistic dualism, the answer is to posit brute psychophysical laws, that is, brute laws that connect functional states of the brain with consciousness. Much like Descartes, who took mind-body interaction to be a brute fact, Chalmers’ dualism takes connections between consciousness and the brain as a brute fact. Following the rationalist dogma, he just denies that those connections hold as a matter of necessity. This shows that the only relevant difference between naturalistic dualism and a hardcore physicalist (e.g., identity theory) is where to locate the brute facts. I don’t claim this is a trivial disagreement— just that it’s ultimately a disagreement about whether phenomenal rationalism is true.

Second, consider contemporary versions of panpsychism²³. This theory holds that consciousness is both fundamental and ubiquitous; that is, consciousness is not grounded in anything else, and it’s all over the place among the fundamental building blocks of reality. This view usually starts from a different philosophical concern, namely, that the real puzzle is the nature of the physical. This is the flipside of the hard problem. On this view, consciousness isn’t explained by the physical— it’s the other way around. Consciousness constitutes the fundamental physical facts. This suggests, for instance, that electrons or quarks possess some form of consciousness. Of course, it wouldn’t be like human consciousness, but it would still count as a phenomenal property (a “micro-experience”). You may wonder: how is this supposed to solve the hard problem of consciousness? Roughly, the idea is that once consciousness is among the basic ingredients of the physical building blocks, it shouldn’t be a mystery why consciousness arises out of physical systems. After all, consciousness is what partially explains the nature of those physical systems.

However, it’s widely agreed that panpsychism faces a problem known as the ‘combination problem’. The worry is that it’s very hard to see how micro-experiences (or proto-experiences) yield human consciousness. Experiences from different subjects just don’t seem to add up— at least there is no intelligible way to combine them. Thus, it seems we have another explanatory gap, and therefore still facing the hard problem. Notice that the issue again is that there’s no intelligible connection (“a priori entailment”) from micro-experiences to human consciousness. This is only a problem if you assume there can’t be brute a priori facts about consciousness. In other words, what’s at stake again is phenomenal rationalism.

I hope to have shown that the question of physicalism isn’t that important since the hard problem arises for non-physicalist accounts of the mind as well. The case of dualism and panpsychism shows that what lies at the heart of the problem of consciousness is ultimately rationalism. Ironically, if you care about the status of physicalism²⁴, given that its main trouble has to do with explanatory gaps, the way forward is to examine rationalism. Again, recall our leitmotif: the bat is the ghost.

4. For the Philosophers: Post-Rationalism

Now it’s time for a positive methodological proposal. This will take the form of a new set of questions. My hope is that these questions will move the discussion into a fruitful direction.

Perhaps the most straightforward and obvious question would be: Is phenomenal rationalism true? Equivalently: Why are brute a priori facts ruled out from a theory of consciousness? Even more specifically: Why can’t we accept brute necessary connections between brain states and conscious states? These are the questions at the core of the hard problem; my positive proposal is that we should prioritize them. We can also break these questions down into more fundamental questions that are in the vicinity. Here are a few for philosophers:

Why can’t a priori facts be brute?

This investigates why facts belonging to the a priori domain (logic, mathematics, necessary beings, and presumably consciousness) must be intelligible. Perhaps within those domains, there are necessary truths that aren’t transparent to our understanding— not even to the understanding of an ideal agent. Even after grasping all the relevant concepts that describe the relevant a priori facts, we still can’t figure out what follows from them. Those truths might be inscrutable.

Why are facts about consciousness a priori facts?

This explores why consciousness should belong to the same domain that typically demands transparent understanding—like logic, mathematics, or God (if such a being exists). Even if we accept that all a priori facts are intelligible, one might deny that facts about consciousness are a priori facts. The thesis that the essence of consciousness is intelligible usually comes from the old Cartesian doctrine in which we have privileged access to our conscious minds or a transparent understanding of its essence. But this isn’t philosophically innocent.

Why does ontological explanation require intelligibility?

This looks into why relations of dependence between facts must be intelligible at all. When we say that biological facts depend on chemical facts, and chemical facts depend on physical facts, we are tracking ontological dependencies that are supposed to be explanatory. The same goes when we say that consciousness depends on brain states. But why should these explanations be intelligible in the rationalist sense? Perhaps such dependencies exist, but it isn’t rationally transparent why they obtain (in the technical jargon: there is no a priori entailment between the relevant facts). The alternative is to have inscrutable relations of dependence, but what do they look like?

Here is an approximate picture that philosophers have already suggested²⁵. In successful cases of reduction (water = H₂O), identities explain correlations by revealing a mistaken assumption in the original question. If someone asks, "Why does the chemical compound H₂O predict water's behavior and properties so well?" it's legitimate to answer: "Because water just is H₂O." Yet there’s no conceptual or a priori connection between water and H₂O. It's simply a brute fact that they're the same kind of thing. Brute identities explain by revealing incorrect presuppositions. Maybe something similar applies to ontological dependencies—consciousness depending on the brain, for instance—such that they explain by eliminating mistaken presuppositions.

What is intelligibility anyway?

I have been assuming — perhaps wrongly— that we all find intelligible what I mean by ‘intelligible’. Even if we do, there is always room for refinement. The pre-theoretical notion of intelligibility is something like available to understanding. But available to whose understanding? There’s also no uncontentious theory of understanding in epistemology. Moreover, the technical approaches to intelligibility usually rest on the idea that facts P make intelligible facts Q if and only if they have a relation of conceptual entailment (i.e., the concepts used to express Q-facts are already contained in the concepts we use to express P-facts) or a priori entailment (i.e., conditionals like ‘if P then Q’ are a priori knowable). But it’s not obvious that these are the only options available. Thus, a theory of intelligibility would help answer the previous questions, understand what an explanatory gap is, and assess whether rationalists’ demands are plausible.

Is phenomenal rationalism the default approach in the Asian philosophical tradition?

Finally, a crucial question for historians of philosophy. When we say that the bat is the ghost, we are saying that the rationalist bias has persisted for centuries. But it’s an open question whether this bias is just a feature of Western thought. Perhaps in Asian philosophical traditions like Indian philosophy or Buddhist philosophy, which have discussed consciousness over millennia, thinkers haven’t been so enthusiastic about rationalism. I don’t know enough to argue for such historical claims, but one natural suggestion is to investigate what these philosophers thought about brute facts, inscrutable truths, and the realm of the unintelligible. Did they, for instance, reject the principle of sufficient reason? I must confess that I wouldn’t be surprised if Buddhist philosophers exorcised the Cartesian ghost of rationalism centuries ago.

5. For the Scientists: Unintelligible Intelligence

I have suggested some questions that philosophers can explore to move forward in the problem of consciousness. I don’t have answers to those questions, but I see no “hard problem” among them. Can the sciences also bring something to the table? I think that our current scientific models can significantly contribute to the question of rationalism, and therefore, to the problem of consciousness. In particular, it appears that our computational models of intelligence are already situated in a post-rationalist paradigm. Thus, we can gain insights for a post-rationalist theory of consciousness from the current revolution in AI.

In earlier sections, I claimed that computer development represented the pinnacle of rationalism because it shows that a mechanical implementation of rationality is possible. But this claim requires qualification. In the field of AI, there have been at least two major traditions. On one hand, we have the symbolic approach, where intelligence arises from the top-down application of abstract rules and logic to knowledge states, transparently guiding system behavior. Such a tradition gave us remarkable systems like Deep Blue, the chess supercomputer, but the approach also faced major issues like the frame problem. On the other hand, we have the statistical approach, where intelligence arises from bottom-up pattern recognition in massive datasets guided by learning algorithms. This approach gave us Large Language Models (LLMs) like GPT and much more to come. Yet one pressing issue that engineers face with these systems is that they are, to a certain extent, black boxes²⁶. The inner workings of deep neural networks aren’t always transparent. Often, they operate through billions of parameters, and their contribution to the final output isn’t understood. Computer scientists are trying to find new statistical methods to reverse-engineer this problem, but for the most part, many systems are still uninterpretable. However, what matters to us is that inscrutability doesn’t prevent these systems from exhibiting intelligence. Simply put, black box AI is unintelligible intelligence.

Why is this relevant to the problem of consciousness? Because this situates black box AI within a post-rationalist philosophy of mind. Deep learning models weren’t the sort of intelligent systems envisioned by the early rationalists. If anything, neural networks are distant descendants of the theories of thought defended by British empiricists like David Hume. I’m aware this parallel is more figurative than accurate, but it still illustrates how computation gives rise to intelligence despite the explanatory gaps. So, considerations from black box AI motivate the following question: If intelligence can arise from unintelligible patterns and networks of information processing, why not consciousness? Notice that early rationalists would have rejected black box intelligence, yet we have it today without any hard problem of intelligence. Phenomenal rationalists reject unintelligible consciousness, but perhaps that’s all there is to consciousness. To dismiss this possibility is to assume phenomenal rationalism right from the start.

One important question remains: how would neuroscience look in the post-rationalist picture? From a scientific standpoint, it’s not too different because the dispute over rationalism isn’t an empirical matter. It’s a philosophical debate about what an ultimate explanation of consciousness looks like. Scientific theories of consciousness that appear on the right track (e.g., the global workspace theory²⁷) are frequently dismissed by rationalists as providing mere neural correlates; they are not explanatory by their standards. What I’ve been trying to do is to question those standards because it’s not as if better neurobiological or computational models would ever satisfy their demands. To the rationalist, reductive or functional explanation is not possible even in principle. Let me show you just how demanding rationalist requirements really are.

Imagine this scenario: 100 years from now, after thousands of experiments and computational simulations, we discover the perfect correlate of consciousness—a functional process found in every creature we recognize as conscious, from human adults to mammals. Call it the 'phenomenal algorithm.' This algorithm is abstract enough that we also find it running in some complex AI systems. Using this algorithm combined with subjective reports, we can predict with unprecedented accuracy how visual consciousness fades after brain damage in blindsight, or how it becomes scrambled under psychedelic drugs. The precision is so remarkable that our algorithm doesn't just tell us whether someone is conscious —it predicts their exact subjective reports. Despite all this evidence, our rationalist would tell us—without hesitation—that since we can imagine this algorithm running in zombie mode (in the absence of consciousness), the two can't be intelligibly connected. And if there's no intelligible connection between them, they can't be identical. The phenomenal algorithm, no matter how perfect, remains just a correlate—nothing more.

Perhaps you still think that rationalism offers a more promising future for the neuroscience of consciousness. But don't be misled by the label. Both views accept that the relations between consciousness and the brain are brute— the only difference is that the rationalist would add: it's because those connections are brute that consciousness and the brain can't be the same (or necessarily depend on each other). Given these radical demands, abandoning rationalism doesn’t change much of the actual scientific practice, but it provides scientists with the philosophical tools to reply to skeptical philosophers and a solid foundation for the study of consciousness. Importantly, post-rationalism doesn’t suggest that neuroscience is finished; it encourages scientists to keep improving their models by dispelling the idea of an unsolvable hard problem.

Am I suggesting that the relation between consciousness and the brain remains a mystery? No, brute facts aren't mysteries— at least not in the philosophical sense. Chomsky famously distinguished between problems and mysteries. The latter are problems whose solution is cognitively closed to us— they are beyond our cognitive reach, like calculus is for dogs. A philosopher once defended the view that the problem of consciousness is a mystery in Chomsky's sense²⁸. This view states that there must be an explanation of why consciousness arises from the brain, but such an explanation is cognitively closed to us. This is an extreme form of rationalism, and it's certainly not the view that I'm considering. In the post-rationalist approach, it's not that there must be an explanation that happens not to be available due to our limited cognitive capacities. Rather, we are entertaining the very possibility that such an explanation —which demands rational intelligibility— simply doesn’t exist; it’s a metaphysical myth.

The payoff of our journey is dissolving the hard problem —or at least, making it easier. When a problem is so stubborn that it’s claimed to be unsolvable, the philosophically responsible thing to do is to scrutinize the assumptions behind such claims. Perhaps it’s just hard because we’ve been framing it incorrectly. Perhaps, as I argue, the bat is the ghost. But the ghost haunts no more.

Bibliography

Block, N. (1978). Troubles with Functionalism. Minnesota Studies in the Philosophy of Science, 9, 261-325.

Block, N. & Stalnaker, R. (1999). Conceptual Analysis, Dualism, and the Explanatory Gap. The Philosophical Review, 108 (1), 1-46.

Chalmers, D. (1995). Facing Up to the Problem of Consciousness. Journal of Consciousness Studies, 2 (3), 200-19.

Chalmers, D. (1996). The Conscious Mind: In Search of a Fundamental Theory. Oxford: Oxford University Press.

Dehaene, S., & Naccache, L. (2001). Towards a cognitive neuroscience of consciousness: basic evidence and a workspace framework. Cognition, 79(1-2), 1–37.

Della Rocca, M. (2003). A Rationalist Manifesto: Spinoza and the Principle of Sufficient Reason. Philosophical Topics, 31(1/2), 75–93.

Descartes, René (1984–91). The Philosophical Writings of Descartes. Vol 1-3., Eds. John Cottingham, Robert Stoothoff, Dugald Murdoch, and Anthony Kenny. Cambridge: Cambridge University Press.

Goff, P. (2017). Consciousness and Fundamental Reality. Oxford: Oxford University Press.

Fine, K. (1994). Essence and Modality: The Second Philosophical Perspectives Lecture. Philosophical Perspectives, 8, 1–16.

Jackson, F. (1982). Epiphenomenal Qualia. The Philosophical Quarterly, 32 (127), 127-136.

Levine, (1983). Materialism and Qualia: The Explanatory Gap. Pacific Philosophical Quarterly, 64 (4), 354-361.

Leibniz, G.W. (1989). Philosophical Essays. Eds. R. Ariew and D. Garber. Hackett Publishing.

McGinn, C. (1989). Can We Solve the Mind--Body Problem? Mind, 98(391), 349–366.

Nagel, T. (1974). What Is It Like to Be a Bat? The Philosophical Review, 83(4), 435–450. 28

Papineau, D. (1998). Mind the Gap. Philosophical Perspectives, 12, 373-388.

Rudin, C (2019). Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. Nature Machine Intelligence, 1, 206–215.

Ryle, G. (1949). The Concept of Mind. Chicago: The University of Chicago Press.

Schaffer, J. (2017). The Ground Between the Gaps. Philosophers' Imprint, 17 (11), 1-26.

Searle, J. R. (1980). Minds, Brains, and Programs. Behavioral and Brain Sciences, 3(3), 417–424.

Spinoza, B. (1984). The Collected Works of Spinoza. V1., Ed. E. Curley. NJ: Princeton University Press.