The Phi Value of Integrated Information Theory Might Not Be Stable Across Small Changes in Neural Connectivity

In learning and in forgetting, the amount of connectivity between your neurons changes. Throughout your life, neurons die and grow. Through all of this, the total amount of conscious experience you have, at least in your alert, attentive moments, seems to stay roughly the same. You don't lose a few neural connections and with it 80% of your consciousness. The richness of our stream of experience is stable across small variations in the connectivity of our neurons -- or so, at least, it is plausible to think.

One of the best known theories of consciousness, Integrated Information Theory, purports to model how much consciousness a neural system has by means of a value, Φ (phi), that is a mathematically complicated measure of how much "integrated information" a system possesses. The higher the Φ, the richer the conscious experience, the lower the Φ, the thinner the experience. Integrated Information Theory is subject to some worrying objections (and here's an objection by me, which I invite you also to regard as worrying). Today I want to highlight a different concern than these: the apparent failure of Φ to be robust to small changes in connectivity.

The Φ of any particular informational network is difficult to calculate, but the IIT website provides a useful tool. You can play around with networked systems of about 4, 5, or 6 nodes (above 6, the computation time to calculate Φ becomes excessive). Prefab systems are available to download, with Φ values from less than 1 to over 15. It's fun!

But there are two things you might notice, once you play around with the tool for a while:

First, it's somewhat hard to create systems with Φ values much above 1. Slap 5 nodes together and connect them any which way, and you're likely to get a Φ value between 0 and 1.

Second, if you tweak the connections of the relatively high-Φ systems, even just a little, or if you change a logical operator from one operation to another (e.g., XOR to AND), you're likely to cut the Φ value by at least half. In other words, the Φ value of these systems is not robust across small changes.

To explore the second point more systematically, I downloaded the "IIT 3.0 Paper Fig. 17 Specialized majority" network which, when all 5 nodes are lit, has a Φ value of 10.7. (A node's being "lit" means it has a starting value of "on" rather than "off".) I then tweaked the network in every way that it was possible to tweak it by changing exactly one feature. (Due to the symmetry of the network, this was less laborious than it sounds.) Turning off any one node reduces Φ to 2.2. Deleting any one node reduces Φ to 1. Deleting one connection, altering its direction (if unidirectional), or changing it from unidirectional to bidirectional or vice versa, always reduces system's Φ to a value ranging from 2.6 to 4.8. Changing the logic function of one node has effects that are sometimes minor and sometimes large: Changing any one node from MAJ to NOR reduces Φ all the way down to 0.4, while changing any one node to MIN increases Φ to 13.0. Overall, most ways of introducing one minimal perturbation into the system reduce Φ by at least half, and some reduce it by over 90%.

To confirm that the "Specialized majority" network was not unusual in this respect, I attempted a similar systematic one-feature tweaking of "CA Paper Fig 3d, Rule 90, 5 nodes". The 5-node Rule 90 network, with all nodes in the default unlit configuration, has a Φ of 15.2. The results of perturbation are similar to the results for the "Specialized majority" network. Light any one node of the rule 90 network and Φ falls to 1.8. Delete any one arrow and Φ also falls to 1.8. Change any one arrow from bidirectional to unidirectional and Φ falls to 4.8. Change the logic of one node and Φ ranges anywhere from a low of 1.8 (RAND, PAR, and >2) to a high of 19.2 (OR).

These two examples, plus what I've seen in my unsystematic tweaking of other prefab networks, plus my observations about the difficulty of casually constructing a five-node system with Φ much over 1, suggest that, in five-node systems at least, having a high Φ value requires highly specific structures that are unstable to minor perturbations. Small tweaks can easily reduce Φ by half or more.

It would be bad for Integrated Information Theory, as a theory of consciousness, if this high degree of instability in systems with high Φ values scales up to large systems, like the brain. The loss of a few neural connections shouldn't make a human being's Φ value crash down by half or more. Our brains are more robust than that. And yet I'm not sure that we should be confident that the mathematics of Φ has the requisite stability in large, high-Φ systems. In the small networks we can measure, at least, it is highly unstable.

ETA November 10:

Several people have suggested to me that Phi will be more stable to small perturbations as the size of the network increases. I could see how that might be the case (which is why I phrased the concluding paragraph as a worry rather than as an positive claim). Now if Phi, like entropy, were dependent in some straightforward way on the small contributions of many elements, that would be likely to be so. But the mathematics of Phi relies heavily on discontinuities and threshold concepts. I exploit this fact in my earlier critique of the Exclusion Postulate, in which I show that a very small change in the environment of a system, without any change interior to the system, could cause that system to instantly fall from arbitrarily high Phi to zero.

If anyone knows of a rigorous, rather than handwavy attempt to show that Phi in large systems is stable over minor perturbations, I would be grateful if you pointed it out!

AI and Consciousness: A Skeptical Overview, forthcoming with Cambridge

Last week I submitted my latest book manuscript to Cambridge University Press (for their "Element" series of books about 100 pages long): AI and Consciousness: A Skeptical Overview -- because you haven't heard nearly enough about AI and consciousness recently, of course! [winky face]

Maybe you'll appreciate my skeptical stance, at odds both with the boosters who anticipate imminent AI consciousness and with the scoffers who pooh-pooh the possibility. Or maybe you'll loathe my skeptical stance but grudgingly accept it against your will, due to the force of my arguments!

I've pasted the introductory chapter below. The full (citable) manuscript version is available here and here.

[AI and Consciousness, title page]

Chapter One: Hills and Fog

1. Experts Do Not Know and You Do Not Know and Society Collectively Does Not and Will Not Know and All Is Fog.

Our most advanced AI systems might soon – within the next five to thirty years – be as richly and meaningfully conscious as ordinary humans, or even more so, capable of genuine feeling, real self-knowledge, and a wide range of sensory, emotional, and cognitive experiences. In some arguably important respects, AI architectures are beginning to resemble the architectures many consciousness scientists associate with conscious systems. Their outward behavior, especially their linguistic behavior, grows ever more humanlike.

Alternatively, claims of imminent AI consciousness might be profoundly mistaken. Their seeming humanlikeness might be a shadow play of empty mimicry. Genuine conscious experience might require something no AI system could possess for the foreseeable future – intricate biological processes, for example, that silicon chips could never replicate.

The thesis of this book is that we don’t know. Moreover and more importantly, we won’t know before we’ve already manufactured thousands or millions of disputably conscious AI systems. Engineering sprints ahead while consciousness science lags. Consciousness scientists – and philosophers, and policy-makers, and the public – are watching AI development disappear over the hill. Soon we will hear a voice shout back to us, “Now I am just as conscious, just as full of experience and feeling, as any human”, and we won’t know whether to believe it. We will need to decide, as individuals and as a society, whether to treat AI systems as conscious, nonconscious, semi-conscious, or incomprehensibly alien, before we have adequate grounds to justify that decision.

The stakes are immense. If near-future AI systems are richly, meaningfully conscious, then they will be our peers, our lovers, our children, our heirs, and possibly the first generation of a posthuman, transhuman, or superhuman future. They will deserve rights, including the right to shape their own development, free from our control and perhaps against our interests.[1] If, instead, future AI systems merely mimic the outward signs of consciousness while remaining as experientially blank as toasters, we face the possibility of mass delusion on an enormous scale. Real human interests and real human lives might be sacrificed for the sake of entities without interests worth the sacrifice. Sham AI “lovers” and “children” might supplant or be prioritized over human lovers and children. Heeding their advice, society might turn a very different direction than it otherwise would.

In this book, I aim to convince you that the experts do not know, and you do not know, and society collectively does not and will not know, and all is fog.

2. Against Obviousness.

Some people think that near-term AI consciousness is obviously impossible. This is an error in adverbio. Near-term AI consciousness might be impossible – but not obviously so.

A sociological argument against obviousness:

Probably the leading scientific theory of consciousness is Global Workspace theory. Its leading advocate is neuroscientist Stanislas Dehaene.[2] In 2017, years before the surge of interest in ChatGPT and other Large Language Models, Dehaene and two collaborators published an article arguing that with a few straightforward tweaks, self-driving cars could be conscious.[3]

Probably the two best-known competitors to Global Workspace theory are Higher Order theory and Integrated Information Theory.[4] (In Chapters Eight and Nine, I’ll provide more detail on these theories.) Perhaps the leading scientific defender of Higher Order theory is Hakwan Lau – one of the coauthors of that 2017 article about potentially conscious cars.[5] Integrated Information Theory is potentially even more liberal about machine consciousness, holding that some current AI systems are already at least a little bit conscious and that we could easily design AI systems with arbitrarily high degrees of consciousness.[6]

David Chalmers, the world’s most influential philosopher of mind, argued in 2023 for about a 25% degree of confidence in AI consciousness within a decade.[7] That same year, a team of prominent philosophers, psychologists, and AI researchers – including eminent computer scientist Yoshua Bengio – concluded that there are “no obvious technological barriers” to creating conscious AI according to a wide range of mainstream scientific views about consciousness.[8] In a 2025 interview, Geoffrey Hinton, another of the world’s most prominent computer scientists, asserted that AI systems are already conscious.[9] Christof Koch, the most influential neuroscientist of consciousness from the 1990s to the early 2010s, has endorsed Integrated Information Theory, including its liberal implications for the pervasiveness of consciousness.[10]

This is a sociological argument: a substantial probability of near-term AI consciousness is a mainstream view among leading experts. They might be wrong, but it’s implausible that they’re obviously wrong – that there’s a simple argument or consideration they’re neglecting which, if pointed out, would or should cause them to collectively slap their foreheads and say, “Of course! How did we miss that?”

What of the converse claim – that AI consciousness is obviously imminent or already here? In my experience, fewer people assert this. But in case you’re tempted in this direction, note that other prominent theorists hold that AI consciousness is a far-distant prospect if it’s possible at all: neuroscientist Anil Seth; philosophers Peter Godfrey-Smith, Ned Block, and John Searle; linguist Emily Bender; and computer scientist Melanie Mitchell.[11] (Chapter Six will discuss thought experiments by Searle, Bender, and Mitchell, and Chapter Ten will discuss biological views of the sort emphasized by Seth, Godfrey-Smith, and Block.) In a 2024 survey of 582 AI researchers, 25% expected AI consciousness within ten years and 70% expected AI consciousness by the year 2100.[12]

If the believers are right, we’re on the brink of creating genuinely conscious machines. If the scoffers are right, those machines will only seem conscious. I assume that this is a substantive disagreement, not just a disagreement about how to apply the term “consciousness” to a perfectly obvious set of phenomena about which everyone agrees. The future well-being of many people (including, perhaps, many AI people) depends on getting this issue right. Unfortunately, we will not know in time.

The rest of this book is flesh on this skeleton. I canvass a variety of structural and functional claims about consciousness, the leading theories of consciousness as applied to AI, and the best known general arguments for and against near-term AI consciousness. None of these claims or arguments takes us far. It’s a morass of uncertainty.

-------------------------------------------

[1] I assume that AI consciousness and AI rights are closely connected: Schwitzgebel 2024, ch. 11, in preparation. For discussion, see Shepherd 2018; Levy 2024.

[2] Dehaene 2014; Mashour et al. 2020.

[3] Dehaene, Lau, and Kouider 2017. For an alternative interpretation of this article as concerning something other than consciousness in its standard “phenomenal” sense, see note 115.

[4] Some Higher Order theories: Rosenthal 2005; Lau 2022; Brown 2025. Integrated Information Theory: Albantakis et al. 2023.

[5] But see Chapter Eight for some qualifications.

[6] See Tononi’s publicly available response to Scott Aaronson’s objections in Aaronson 2014. However, advocates of IIT also suggest that the most common current computer architectures are unlikely to achieve much consciousness and that consciousness will tend to appear in subsystems of the computer rather than at the level of the computer itself (Findlay et al. 2024/2025).

[7] Chalmers 2023.

[8] Butlin et al. 2023. (I am among the nineteen authors.)

[9] Heren 2025.

[10] Tononi and Koch 2015.

[11] Seth forthcoming; Godfrey-Smith 2024; Block forthcoming; Searle 1980, 1992; Bender 2025; Mitchell 2021.

[12] Dreksler et al. 2025.

Tononi's Exclusion Postulate Would Make Consciousness (Nearly) Irrelevant

One of the most prominent theories of consciousness is Guilio Tononi's Integrated Information Theory. The theory is elegant and interesting, if a bit strange. Strangeness is not necessarily a defeater if, as I argue, something strange must be true about consciousness. One of its stranger features is what Tononi calls the Exclusion Postulate. The Exclusion Postulate appears to render the presence or absence of consciousness almost irrelevant to a system's behavior.

Here's one statement of the Exclusion Postulate:

The conceptual structure specified by the system must be singular: the one that is maximally irreducible (Φ max). That is, there can be no superposition of conceptual structures over elements and spatio-temporal grain. The system of mechanisms that generates a maximally irreducible conceptual structure is called a complex... complexes cannot overlap (Tononi & Koch 2014, p. 5).

The basic idea here is that conscious systems cannot nest or overlap. Whenever two information-integrating systems share any parts, consciousness attaches to the one that is the most informationally integrated, and the other system is not conscious -- and this applies regardless of temporal grain.

The principle is appealing in a certain way. There seem to be lots of information-integrating subsystems in the human brain; if we deny exclusion, we face the possibility that the human mind contains many different nesting and overlapping conscious streams. (And we can tell by introspection that this is not so -- or can we?) Also, groups of people integrate information in social networks, and it seems bizarre to suppose that groups of people might have conscious experience over and above the individual conscious experiences of the members of the groups (though see my recent work on the possibility that the United States is conscious). So the Exclusion Postulate allows Integrated Information Theory to dodge what might otherwise be some strange-seeming implications. But I'd suggest that there is a major price to pay: the near epiphenomenality of consciousness.

Consider an electoral system that works like this: On Day 0, ten million people vote yes/no on 20 different ballot measures. On Day 1, each of those ten million people gets the breakdown of exactly how many people voted yes on each measure. If we want to keep the system running, we can have a new election every day and individual voters can be influenced in their Day N+1 votes by the Day N results (via their own internal information integrating systems, which are subparts of the larger social system). Surely this is society-level information integration if anything is. Now according to the Exclusion Postulate, whether the individual people are conscious or instead the societal system is conscious will depend on how much information is integrated at the person level vs. the societal level. Since "greater than" is sharply dichotomous, there must be an exact point at which societal-level information integration exceeds the person-level information integration. Tononi and Koch appear to accept a version of this idea in 2014, endnote xii [draft of 26 May 2014]. As soon as this crucial point is reached, all the individual people in the system will suddenly lose consciousness. However, there is no reason to think that this sudden loss of consciousness would have any appreciable effect on their behavior. All their interior networks and local outputs might continue to operate in virtually the same way, locally inputting and outputting very much as before. The only difference might be that individual people hear back about X+1 votes on the Y ballot measures instead of X votes. (X and Y here can be arbitrarily large, to ensure sufficient informational flow between individuals and the system as a whole. We can also allow individuals to share opinions via widely-read social networks, if that increases information integration.) Tononi offers no reason to think that a small threshold-crossing increase in the amount of integrated information (Φ) at the societal level would profoundly influence the lower-level behavior of individuals. Φ is just a summary number that falls out mathematically from the behavioral interactions of the individual nodes in the network; it is not some additional thing with direct causal power to affect the behavior of those nodes.

I can make the point more vivid. Suppose that the highest-level Φ in the system belongs to Jamie. Jamie has a Φ of X. The societal system as a whole has a Φ of X-1. The highest-Φ individual person other than Jamie has a Φ of X-2. Because Jamie's Φ is higher than the societal system's, the societal system is not a conscious complex. Because the societal system is not a conscious complex, all those other individual people with Φ of X-2 or less can be conscious without violating the Exclusion Postulate. But Tononi holds that a person's Φ can vary over the course of the day -- declining in sleep, for example. So suppose Jamie goes to sleep. Now the societal system has the highest Φ and no individual human being in the system is conscious. Now Jamie wakes and suddenly everyone is conscious again! This might happen even if most or all of the people in the society have no knowledge of whether Jamie is asleep or awake and exhibit no changes in their behavior, including in their self-reports of consciousness.

More abstractly, if you are familiar with Tononi's node-network pictures, imagine two very similar largish systems, both containing a largish subsystem. In one of the two systems, the Φ of the whole system is slightly less than that of the subsystem. In the other, the Φ of the whole system is slightly more. The node-by-node input-output functioning of the subsystem might be virtually identical in the two cases, but in the first case, it would have consciousness -- maybe even a huge amount of consciousness if it's large and well-integrated enough! -- and in the other case it would have none at all. So its consciousness or lack thereof would be virtually irrelevant to its functioning.

It doesn't seem to me that this is a result that Tononi would or should want. If Tononi wants consciousness to matter, given the Exclusion Postulate, he needs to show why slight changes of Φ, up or down at the higher level, would reliably cause major changes in the behavior of the subsystems whenever the Φ(max) threshold is crossed at the higher level. There seems to be no mechanism that ensures this.

Disunity and Indeterminacy in Artificial Consciousness (and Maybe in Human Consciousness Too)

Our understanding of the nature of consciousness derives mainly from our understanding of the nature of consciousness in our favorite animal (us, of course). But the features of consciousness in our favorite animal might be specific to that animal rather than universal.

Let's consider two such features and whether we should expect them in conscious AI systems, if conscious AI systems are ever possible.

Unity: Our conscious experiences at any given moment are bound together into a single unified experience, rather than transpiring in separate streams. If I'm sitting on a wet park bench, I might (a.) visually experience the leafy green trees around me, (b.) tactilely experience the cold dampness soaking into my jeans, and (c.) consciously recall the smaller trees of yesteryear. Normally -- perhaps necessarily -- three such experiences would not run in disconnected streams. They would join into a composite experience of (a)-with-(b)-with-(c). I experience not just trees, cold dampness, and a memory of yesteryear, but all three together as a unified bundle.

Determinacy: At any given moment, I am either determinately conscious or determinately nonconscious (as in anesthesia or dreamless sleep). Likewise, I either determinately do, or determinately do not, have any particular experience. Gray-area cases are at least unusual and maybe impossible. Even the simplest, barest cases are still determinate. Consider visual experience: We might imagine the visual field narrowing and losing content until only a gray dot remains -- and then the dot winks out. That dot, however minimal, is still determinately experienced. When it winks out, consciousness determinately disappears. There is no half-winked state between the minimal gray dot and complete absence of visual experience.

My thought is that we should not expect unity and determinacy to be general features of conscious AI systems (if conscious AI is possible). To see why, let's start by assuming the Global Workspace Theory of consciousness. I focus on Global Workspace Theory because it's probably the leading scientific theory of consciousness and because its standard formulation (Dehaene's version) invites the assumption of unity and determinacy.

Global Workspace Theory divides the mind into local information processing modules linked by a shared global workspace. Information becomes conscious when it is broadcast into the workspace. Suppose your auditory system registers the faint honk of a distant car horn. You're absorbed in reading philosophy and accustomed to ignoring traffic noise, so this representation isn't selected for further processing. It's not a target of attention, not broadcast into the workspace, and not consciously experienced. (If you think you constantly consciously experience background sounds, you can't hold a standard Global Workspace view.) Once you attend to the noise, for whatever reason, that information "ignites" into the global workspace, becoming available to a wide variety of "downstream" processes: You can think about it, plan around it, verbally report it, store it in long-term memory, and flexibly combine it with other information in the workspace. On Global Workspace Theory, being available in this way just is what it is for the information to be consciously experienced.

This model suggests unity and determinacy. Since there is just one global workspace, and since that workspace enables flexible integration of everything it contains, it makes sense that its various elements will combine into a unified experience. And on Dehaene's version, ignition into the workspace is a sharp-boundaried event: Information either completely ignites, becoming available for all downstream processes, or it does not. There is no (or only rarely) partial ignition. This can explain determinacy.

But future AI systems might not share this structure. They might have multiple or partially overlapping workspaces. Different specialized subsystems might have access to different regions of a partly-shared workspace. Some animals, such as snails and octopuses, distribute processing among multiple ganglia or neural centers that are less tightly coupled than the hemispheres of the human brain. A robot might broadcast information relevant to locomotion to one area and information relevant to speech to another with limited connectivity.

If the subsystems are entirely disconnected, the result might be entirely discrete centers of subjective experience within a single organism or machine. But if they are partly connected, experience might be only partly unified. In the park bench example, the experience of the trees might be unified with the experience of dampness, and the experience of dampness with memories of yesteryear, but the experience of the trees might not be unified with the memories. (Unification would not then be a transitive relation.) Alternatively, some weaker relation of partial unification might hold among the visual, tactile, and memorial experiences. If this seems inconceivable or impossible, see Sophie Nelson's and my article on indeterminate or fractional subjects.

More abstractly: There's no compelling architectural reason why an AI system would have to make information available either to all downstream processes or to none. A workspace defined in terms of downstream availability could be a patchwork of partial availabilities rather than a fully global all-or-nothing broadcast.

For the same reason, ignition into the workspace needn't be all-or-nothing. Between full ignition with determinate consciousness and no ignition with determinate nonconsciousness, there might be in-between, gray-area half-ignitions that are neither determinately conscious nor determinately nonconscious. Nearly every property with a complex physical or functional basis allows indeterminate, borderline cases: baldness, extraversion, greenness, happiness, whether you're wearing a shoe, whether a country is a democracy. The human global workspace might minimize indeterminacy -- like it's rarely indeterminate in basketball whether the ball has gone through the hoop. But change the architecture and indeterminacy might become common: a half-hearted ignition, or just enough information-sharing to make it indeterminate whether a workspace even exists. (If indeterminacy about consciousness strikes you as inconceivable or impossible, see my 2023 article on borderline consciousness.)

Global Workspace Theory might of course be wrong. But most other theories of consciousness make my argument at least as easy. Dennett's fame-in-the-brain version of broadcast theory explicitly permits disunity and indeterminacy. Higher Order Theories admit the same fragmentation and, probably, gradualism. So do biological theories and theories that focus on embodiment. (Integrated Information Theory is an exception: Its axioms require bright-lined unity and determinacy. But as I've argued, those bright-line axioms lead to unpalatable consequences.)

Recognizing these possibilities for AI systems invites the further thought: Maybe we humans aren't quite as unified as we normally suppose. Maybe indeterminate and disunified consciousness is common. Maybe processes outside of attention hover indeterminately between being conscious and nonconscious. Maybe some processes are only partly unified. If it seems otherwise in introspection and memory, maybe that's because introspection and memory tend to impose unity and determinacy where none was before.

[a Paul Klee painting, untitled 1914: source]

Two Problems with Extending Theories of Consciousness to the Case of the Garden Snail

In an earlier post, I argued that the question “is there something it’s like to be a garden snail?” or equivalently “are garden snails conscious?” admits of three possible answers – yes, no, and *gong* (that is, neither yes nor no) – and that each of these answers has some antecedent plausibility. That is, prior to detailed theoretical argument, all three answers should be regarded as viable possibilities (even if we have a favorite). To settle the question, then, we need a good theoretical argument that would reasonably convince people who are antecedently attracted to a different view.

It is difficult to see how such an argument could go, for two related reasons: (1.) lack of sufficient theoretical common ground and (2.) the species-specificity of introspective and verbal evidence.

---------------------------------------

Lack of sufficient theoretical common ground.

Existing theories of consciousness, by leading researchers, range over practically the whole space of possibilities from panpsychism on one end, according to which consciousness is ubiquitous, to very restrictive meta-representational views on the other end that deny consciousness even to dogs.

The most common (which is not to say the best) arguments against these extreme views illustrate the common ground problem. The most common argument against panpsychism -- the reason most people reject it, I suspect -- is just that it seems absurd to suppose that consciousness is literally everywhere, even in, say, protons or simple logic gates. We know, we think, prior to our theory-building, that the range of conscious entities does not include protons or simple logic gates! Some of us -- including those who become panpsychists -- might hold that commitment only lightly, ready to abandon it if presented attractive theoretical arguments to the contrary. However, many of us strongly prefer more moderate views. We feel, not unreasonably, more confident that there is nothing it is like to be a proton than we could ever be that a clever philosophical argument to the contrary was in fact sound. Thus, we construct and accept our moderate views of consciousness partly from the starting background assumption that consciousness isn’t that abundant. If a theory looks like it implies that protons are conscious, we reject the theory rather than accepting the implication; and no doubt we can find some dubious-enough step in the panpsychist argument if we are motivated to do so.

Similarly, the most common argument against extremely sparse views that deny consciousness to dogs and babies is that it seems absurd to suppose that dogs and babies are not conscious. We know, we think, prior to our theory-building, that the range of conscious entities includes dogs and babies. Thus, we construct and accept our moderate views of consciousness partly on the starting background assumption that consciousness isn’t that sparse.

In order to develop a general theory of consciousness, one needs to make some initial assumptions about the approximate prevalence of consciousness. Some theories, from the start, will be plainly liberal in their implications about the abundance of consciousness. Others will be plainly conservative. Such theories will rightly be unattractive to people whose initial assumptions are very different; and if those initial assumptions are sufficiently strongly held, theoretical arguments with the type of at-best-moderate force that we normally see in the philosophy and psychology of consciousness will be insufficiently strong to reasonably dislodge those initial assumptions.

For example, Integrated Information Theory is a lovely theory of consciousness. Well, maybe it has a few problems, but it is renowned, and it has a certain elegance. It is also very nearly panpsychist, holding that consciousness is present wherever information is integrated, even in tiny little systems with simple connectivity, like simple logic gates. For a reader who enters the debates about consciousness attracted to the idea that consciousness might be sparsely distributed in the universe, it’s hard to imagine any sort of foreseeably attainable evidence that ought rightly to lead them to reject that sparse view in favor of a view so close to panpsychism. They might love IIT, but they could reasonably regard it as a theory of something other than conscious experience – a valuable mathematical measure of information integration, for example.

Or consider a moderate view, articulated by Zohar Bronfman, Simona Ginsburg, and Eva Jablonka. Bronfman and colleagues generate a list of features of consciousness previously identified by consciousness theorists, including “flexible value systems and goals”, “sensory binding leading to the formation of a compound stimulus”, a “representation of [the entity’s] body as distinct from the external world, yet embedded in it”, and several other features (p. 2). It’s an intriguing idea. Determining the universal features of consciousness and then looking for a measureable functional relationship that reliably accompanies that set of features -- theoretically, I can see how that is a very attractive move. But why those features? Perhaps they are universal to the human case (though even that is not clear), but it’s doubtful that someone antecedently attracted to a more liberal theory is likely to agree that flexible value systems are necessary for low-grade consciousness. If you like snails... well, why not think they have integration enough, learning enough, flexibility enough? Bronfman and colleagues’ criteria are more stipulated than argued for.

---------------------------------------

The species-specificity of verbal and introspective evidence.

The study of consciousness appears to rely, partly, but in an important way, on researchers’ or participants’ introspections, judgments about their experiences, or verbal reports, which need somehow to be related to physical or functional processes. We know about dream experiences, or inner speech, or visual imagery, or the presence or absence of an experience of unattended phenomena in our perceptual fields, partly because of what people judge or say about their experiences. Despite disagreements about ontology and method, this appears to be broadly accepted among theorists of consciousness.

Behavior and physiology are directly observable (or close enough), but the presence or absence of consciousness must normally be inferred -- or at least this is so once we move beyond the most familiar cases of intuitive consensus. However, the evidential base grounding such inferences is limited. The farther we move away from the familiar human case, the shakier our ground. We have to extrapolate in a risky way, far beyond the scope of our direct introspective and verbal evidence. Perhaps an argument for extrapolation to nearby species (apes? all mammals? all vertebrates?) can be made on grounds of evolutionary continuity and morphological similarity. Extrapolating beyond the familiar cases to, for example, garden snails will inevitably be conjectural and uncertain. The uncertainties involved provide basis for ample reasonable doubt among theorists who are antecedently attracted to very different views.

Let’s optimistically suppose that we learn that, in humans, consciousness involves X, Y, and Z physiological or functional features. Now, in snails we see X’, Y’, and Z’, or maybe W and Z”. Are X’, Y’, and Z’, or W and Z”, close enough? Maybe consciousness in humans requires recurrent neural loops of a certain sort (Humphrey 2011; Lamme 2018). Well, snail brains have some recurrent processing too. But of course it doesn’t look either entirely like the recurrent processing that we see in the human case when we are conscious, nor entirely like the recurrent processing that we see in the human case when we’re not conscious. Or maybe consciousness involves availability to, or presence in, working memory or a “global workspace” (Baars 1988; Dehaene and Changeux 2011; Prinz 2012). Well, information travels broadly through snail brains, enabling coordinated action. Is that global workspace enough? It’s like our workspace in some ways, unlike it in others. In the human case, we might be able to -- if things go very well! -- rely on introspective reports to help ground a theory about how broadly information must be shared within our cognitive system for that information to be consciously experienced, but it is by no means clear how we should then generalize such findings to the case of the garden snail.

So we can imagine that the snail is conscious, extrapolating from the human case on grounds of properties we share with the snail; or we can imagine that the snail is not conscious, extrapolating from the human case on grounds of properties we don’t share with the snail. Both ways of doing it seem defensible, and we can construct attractive, non-empirically-falsified theories that deliver either conclusion. We can also think, again with some plausibility, that the presence of some relevant properties and the lack of other relevant properties makes it a case where the human concept of consciousness fails to determinately apply.

---------------------------------------

[image source]

Does Global Workspace Theory Solve the Question of AI Consciousness?

Hint: no.

Below are three sections from Chapter Eight of my manuscript in draft, AI and Consciousness, fresh new version available today here. Comments welcome!

[image adapted from Dehaene et al. 2011]

1. Global Workspace Theories and Access.

The core idea of Global Workspace Theory is simple. Sophisticated cognitive systems like the human mind employ specialized processes that operate to a substantial extent in isolation. We can call these modules, without committing to any strict interpretation of that term.[1] For example, when you hear speech in a familiar language, some cognitive process converts the incoming auditory stimulus into recognizable speech. When you type on a keyboard, motor functions convert your intention to type a word like “consciousness” into nerve signals that guide your fingers. When you try to recall ancient Chinese philosophers, some cognitive process pulls that information from memory without (amazingly) clogging your consciousness with irrelevant information about German philosophers, British prime ministers, rock bands, or dog breeds.

Of course, not all processes are isolated. Some information is widely shared, influencing or available to influence many other processes. Once I recall the name “Zhuangzi”, the thought “Zhuangzi was an ancient Chinese philosopher” cascades downstream. I might say it aloud, type it out, use it as a premise in an inference, form a visual image of Zhuangzi, contemplate his main ideas, attempt to sear it into memory for an exam, or use it as a clue to decipher a handwritten note. To say that some information is in “the global workspace” just is to say that it is available to influence a wide range of cognitive processes. According to Global Workspace Theory, a representation, thought, or cognitive process is conscious if and only if it is in the global workspace – if it is “widely broadcast to other processors in the brain”, allowing integration both in the moment and over time.[2]

Recall the ten possibly essential features of consciousness from Chapter Three: luminosity, subjectivity, unity, access, intentionality, flexible integration, determinacy, wonderfulness, specious presence, and privacy. [Blog readers: You won't have read Chapter Three, but try to ride with it anyway.] Global Workspace Theory treats access as the central essential feature.

Global Workspace theory can potentially explain other possibly essential features. Luminosity follows if processes or representations in the workspace are available for introspective processes of self-report. Unity might follow if there’s only one workspace, so that everything in it is present together. Determinacy might follow if there’s a bright line between being in the workspace and not being in it. Flexible integration might follow if the workspace functions to flexibly combine representations or processes from across the mind. Privacy follows if only you can have direct access to the contents of your workspace. Specious presence might follow if representations or processes generally occupy the workspace for some hundreds of milliseconds.

In ordinary adult humans, typical examples of conscious experience – your visual experience of this text, your emotional experience of fear in a dangerous situation, your silent inner speech, your conscious visual imagery, your felt pains – appear to have the broad cognitive influences Global Workspace Theory describes. It’s not as though we commonly experience pain but find that we can’t report it or act on its basis, or that we experience a visual image of a giraffe but can’t engage in further thinking about the content of that image. Such general facts, plus the theory’s potential to explain features such as luminosity, unity, determinacy, flexible integration, privacy, and specious presence, lend Global Workspace Theories substantial initial attractiveness.

I have treated Global Workspace Theory as if it were a single theory, but it encompasses a family of theories that differ in detail, including “broadcast” and “fame” theories – any theory that treats the broad accessibility of a representation, thought, or process as the central essential feature making it conscious.[3]

Consider two contrasting views: Dehaene’s Global Neuronal Workspace Theory and Daniel Dennett’s “fame in the brain” view. Dehaene holds that entry into the workspace is all-or-nothing. Once a process “ignites” into the workspace, it does so completely. Every representation or process either stops short of entering consciousness or is broadcast to all available downstream processes. Dennett’s fame view, in contrast, admits degrees. Representations or processes might be more or less famous, available to influence some downstream cognitive processes without being available to influence others. There is no one workspace, but a pandemonium of competing processes.[4] If Dennett is correct, luminosity, determinacy, unity, and flexible integration all potentially come under threat in a way they do not as obviously come under threat on Dehaene’s view.[5]

Dennettian concerns notwithstanding, all-or-nothing ignition into a single, unified workspace is currently the dominant version of Global Workspace Theory. The issue remains unsettled and has obvious implications for the types of architectures that might plausibly host AI consciousness.

2. Consciousness Outside the Workspace; Nonconsciousness Within It?

Global Workspace Theory is not the correct theory of consciousness unless all and only thoughts, representations, or processes in the Global Workspace are conscious. Otherwise, something else, or something additional, is necessary for consciousness.

It is not clear that even in ordinary adult humans a process must be in the Global Workspace to be conscious. Consider the case of peripheral experience. Some theorists maintain that people have rich sensory experiences outside of focal attention: a constant background experience of your feet in your shoes and objects in the visual periphery.[6] Others – including Global Workspace theorists – dispute this. Introspective reports vary, and resolving such issues is methodologically tricky.

One methodological problem: People who report constant peripheral experiences might mistakenly assume that such experiences are always present because they are always present whenever they think to check, and the very act of checking might generate those experiences. This is sometimes called the “refrigerator light illusion”, akin to the error of thinking the refrigerator light is always on because it’s always on when you open the door to check.[7] On this view, you’re only tempted to think you have constant tactile experience of your feet in your shoes because you have that experience on those rare occasions when you’re thinking about whether you have it. Even if you now seem to have a broad range of experiences in different sensory modalities simultaneously, this could result from an unusual act of dispersed attention, or from “gist” perception or “ensemble” perception, in which you are conscious of the general gist or general features of a scene, knowing that there are details, without actually experiencing those unattended details.[8]

The opposite mistake is also possible. Those who deny a constant stream of peripheral experiences might simply be failing to notice or remember them. The fact that you don’t remember now the sensation of your feet in your shoes two minutes ago hardly establishes that you lacked the sensation at the time. Although many people find it introspectively compelling that their experience is rich with detail or that it is not, the issue is methodologically complex because introspection and memory are not independent of the phenomena to be observed.[9]

If we do have rich sensory experience outside of attention, it is unlikely that all of that experience is present in or broadcast to a Global Workspace. Unattended peripheral information is rarely remembered or consciously acted upon, tending to exert limited downstream influence – the paradigm of information that is not widely broadcast. Moreover, the Global Workspace is generally characterized as limited capacity, containing only a few thoughts, representations, objects, or processes at a time – those that survive some competition or attentional selection – not a welter of richly detailed experiences in many modalities at once.[10]

A less common but equally important objection runs in the opposite direction: Perhaps not everything in the Global Workspace is conscious. Some thoughts, representations, or processes might be widely broadcast, shaping diverse processes, without ever reaching explicit awareness.[11] Implicit racist assumptions, for example, might influence your mood, actions, facial expressions, and verbal expressions. The goal of impressing your colleagues during a talk might have pervasive downstream effects without occupying your conscious experience moment to moment.

The Global Workspace theorist who wants to allow that such processes are not conscious might suggest that, at least for adult humans, processes in the workspace are generally also available for introspection. But there’s substantial empirical risk in this move. If the correlation between introspective access and availability for other types of downstream cognition isn’t excellent, the Global Workspace theorist faces a dilemma. Either allow many conscious but nonintrospectable processes, violating widespread assumptions about luminosity, or redefine the workspace in terms of introspectability, which amounts to shifting to a Higher Order view.

3. Generalizing Beyond Vertebrates.

The empirical questions are difficult even in ordinary adult humans. But our topic isn’t ordinary adult humans – it’s AI systems. For Global Workspace Theory to deliver the right answers about AI consciousness, it must be a universal theory applicable everywhere, not just a theory of how consciousness works in adult humans, vertebrates, or even all animals.

If there were a sound conceptual argument for Global Workspace Theory, then we could know the theory to be universally true of all conscious entities. Empirical evidence would be unnecessary. It would be as inevitably true as that rectangles have four sides. But as I argued in Chapter Four, conceptual arguments for the essentiality of any of the ten possibly essential features are unlikely to succeed – and a conceptual argument for Global Workspace Theory would be tantamount to a conceptual argument for the essentiality of access, one of those ten features. Not only do the general observations of Chapter Four suggest against a conceptual guarantee, so also does the apparent conceivability, as described in Section 2 above, of consciousness outside the workspace or nonconsciousness within it – even if such claims are empirically false.

If Global Workspace Theory is the correct universal theory of consciousness applying to all possible entities, an empirical argument must establish that fact. But it’s hard to see how such an empirical argument could proceed. We face another version of the Problem of the Narrow Evidence Base. Even if we establish that in ordinary humans, or even in all vertebrates, a thought, representation, or process is conscious if and only if it occupies a Global Workspace, what besides a conceptual argument would justify treating this as a universal truth that holds among all possible conscious systems?

Consider some alternative architectures. The cognitive processes and neural systems of octopuses, for example, are distributed across their bodies, often operating substantially independently rather than reliably converging into a shared center.[12] AI systems certainly can be, indeed often are, similarly decentralized. Imagine coupling such disunity with the capacity for self-report – an animal or AI system with processes that are reportable but poorly integrated with other processes. If we assume Global Workspace Theory at the outset, we can conclude that only sufficiently integrated processes are conscious. But if we don’t assume Global Workspace Theory at the outset, it’s difficult to imagine what near-future evidence could establish that fact beyond a reasonable standard of doubt to a researcher who is initially drawn to a different theory.

If the simplest version of Global Workspace Theory is correct, we can easily create a conscious machine. This is what Dehaene and collaborators envision in the 2017 paper I discussed in Chapter One. Simply create a machine – such as an autonomous vehicle – with several input modules, several output modules, a memory store, and a central hub for access and integration across the modules. Consciousness follows. If this seems doubtful to you, then you cannot straightforwardly accept the simplest version of Global Workspace Theory.[13]

We can apply Global Workspace Theory to settle the question of AI consciousness only if we know the theory to be true either on conceptual grounds or because it is empirically well established as the correct universal theory of consciousness applicable to all types of entity. Despite the substantial appeal of Global Workspace Theory, we cannot know it to be true by either route.

-------------------------------------

[1] Full Fodorian (1983) modularity is not required.

[2] Mashour et al. 2020, p. 776-777.

[3] E.g., Baars 1988; Dennett 1991, 2005; Tye 2000; Prinz 2012; Dehaene 2014; Mashour et al. 2020.

[4] Whether Dennett’s view is more plausible than Dehaene’s turns on whether, or how commonly, representations or processes are partly famous. Some visual illusions, for example, seem to affect verbal report but not grip aperture: We say that X looks smaller than Y, but when we reach toward X and Y we open our fingers to the same extent, accurately reflecting that X and Y are the same size. The fingers sometimes know what the mouth does not. (Aglioti et al. 1995; Smeets et al. 2020). We adjust our posture while walking and standing in response to many sources of information that are not fully reportable, suggesting wide integration but not full accessibility (Peterka 2018; Shanbhag 2023). Swift, skillful activity in sports, in handling tools, and in understanding jokes also appears to require integrating diverse sources of information, which might not be fully integrated or reportable (Christensen et al. 2019; Vauclin et al. 2023; Horgan and Potrč 2010). In response, the all-or-nothing “ignition” view can explain away such cases of seeming intermediacy or disunity as atypical (it needn’t commit to 100% exceptionless ignition with no gray-area cases), by allowing some nonconscious communication among modules (which needn’t be entirely informationally isolated), and/or by allowing for erroneous or incomplete introspective report (maybe some conscious experiences are too brief, complex, or subtle for people to confidently report experiencing them).

[5] Despite developing a theory of consciousness, Dennett (2016) endorsed “illusionism”, which rejects the reality of phenomenal consciousness (see especially Frankish 2016). I interpret the dispute between illusionists and nonillusionists as a verbal dispute about whether the specific philosophical concept of “phenomenal consciousness” requires immateriality, irreducibility, perfect introspectibility, or some other dubious property, or whether the term can be “innocently” used without invoking such dubious properties. See Schwitzgebel 2016, 2025.

[6] Reviewed in Schwitzgebel 2011, ch. 6; and though limited only to stimuli near the center of the visual field, see the large literature on “overflow” in response to Block 2007.

[7] Thomas 1999.

[8] Oliva and Terralba 2006; Whitney and Leib 2018.

[9] Schwitzgebel 2007 explores the methodological challenges in detail.

[10] E.g., Dehaene 2014; Mashour et al. 2020.

[11] E.g., Searle 1983, ch. 5; Bargh and Morsella 2008; Lau 2022; Michel et al. 2025; see also note 4.

[12] Godfrey-Smith 2016; Carls-Diamante 2022.

[13] See also Goldstein and Kirk-Giannini (forthcoming) for an extended application of Global Workspace Theory to AI consciousness. One might alternatively read Dehaene, Lau, and Kouider 2017 purely as a conceptual argument: If all we mean by “conscious” is “accessible in a Global Workspace”, then building a system of this sort suffices for building a conscious entity. The difficulty then arises in moving from that stipulative conceptual claim to the interesting, substantive claim about phenomenal consciousness in the standard sense described in Chapter Two. Similar remarks apply to the Higher Order aspect of that article. One challenge for this deflationary interpretation is that in related works (Dehaene 2014; Lau 2022) the authors treat their accounts as accounts of phenomenal consciousness. The article concludes by emphasizing that in humans “subjective experience coheres with possession” of the functional features they identify. A further complication: Lau later says that the way he expressed his view in this 2017 article was “unsatisfactory”: Lau 2022, p. 168.

Conscious Subjects Needn't Be Determinately Countable: Generalizing Dennett's Fame in the Brain

It is, I suspect, an accident of vertebrate biology that conscious subjects typically come in neat, determinate bundles -- one per vertebrate body, with no overlap.  Things might be very different with less neurophysiologically unified octopuses, garden snails, split-brain patients, craniopagus twins, hypothetical conscious computer systems, and maybe some people with "multiple personality" or dissociative identity.

Consider whether the following two principles are true:

Transitivity of Unity: If experience A and experience B are each part of the conscious experience of a single subject at a single time, and if experience B and experience C are each part of the conscious experience of a single subject at a single time, then experience A and experience C are each part of the conscious experience of a single subject at a single time.

Discrete Countability: Except in marginal cases at spatial and temporal boundaries (e.g., someone crossing a threshold into a room), in any spatiotemporal region the number of conscious subjects is always a whole number (0, 1, 2, 3, 4...) -- never a fraction, a negative number, an imaginary number, an indeterminate number, etc.

Leading scientific theories of consciousness, such as Global Workspace Theory and Integrated Information Theory are architecturally committed to neat bundles satisfying transitivity of unity and discrete countability.  Global Workspace Theories treat processes as conscious if they are available to, or represented in, "the" global workspace (one per conscious animal).  Integrated Information Theory contains an "exclusion postulate" according to which conscious systems cannot nest or overlap, and has no way to model partial subjects or indiscrete systems.  Most philosophical accounts of the "unity of consciousness" (e.g. Bayne 2010) also invite commitment to these two theses.

In contrast, Dennett's "fame in the brain" model of consciousness -- though a close kin to global workspace views -- is compatible with denying transitivity of unity and discrete countability.  In Dennett's model, a cognitive process or content is conscious if it is sufficiently "famous" or influential among other cognitive processes.  For example, if you're paying close attention to a sharp pain in your toe, the pain process will influence your verbal reports ("that hurts!"), your practical reasoning ("I'd better not kick the wall again"), your planned movements (you'll hobble to protect it), and so on; and conversely, if a slight movement in peripheral vision causes a bit of a response in your visual areas, but you don't and wouldn't report it, act on it, think about it, or do anything differently as a result, it is nonconscious.  Fame comes in degrees.  Something can be famous to different extents among different groups.  And there needn't even be a determinately best way of clustering and counting groups.

[Dall-E's interpretation of a "brain with many processes, some of which are famous"]

Here's a simple model of degrees of fame:

Imagine a million people.  Each person has a unique identifier (a number 1-1,000,000), a current state (say, a "temperature" from -10 to +10), and the capacity to represent the states of ten other people (ten ordered pairs, each containing the identifier and temperature of one other person).

If there is one person whose state is represented in every other person, then that person is maximally famous (a fame score of 999,999).  If there is one person whose state is represented in no other person, that that person has zero fame.  Between these extremes is of course a smooth gradation of cases.

If we analogize to cognitive processes we might imagine the pain in the toe or the flicker in the periphery being just a little famous: Maybe the pain can affect motor planning but not speech, causes a facial expression but doesn't influence the stream of thought you're having about lunch.  Maybe the flicker guides a glance and causes a spike of anxiety but has no further downstream effects.  Maybe they're briefly reportable but not actually reported, and they have no impact on medium- or long-term memory, or they affect some sorts of memory but not others.

The "ignition" claim of global workspace theory is the empirically defensible (but not decisively established) assertion that there are few such cases of partial fame: Either a cognitive process has very limited effects outside of its functional region or it "ignites" across the whole brain, becoming widely accessible to the full range of influenceable processes.  The fame-in-the-brain model enables a different way of thinking that might apply to a wider range of cognitive architectures.

#

We might also extend the fame model to issues of unity and the individuation of conscious subjects.

Start with a simple case: the same setup as before, but with two million people and the following constraint: Processes numbered 1 to 1,000,000 can only represent the states of other processes in that same group of 1 to 1,000,000; and processes numbered 1,000,001 to 2,000,000 can only represent the states of other processes in that group.  The fame groups are disjoint, as if on different planets.  Adapted to the case of experiences: Only you can feel your pain and see your peripheral flicker (if anyone does), and only I can feel my pain and see my peripheral flicker (if anyone does).

This disjointedness is what makes the two conscious subjects distinct from each other.  But of course, we can imagine less disjointedness.  If we eliminate disjointedness entirely, so that processes numbered 1 to 2,000,000 can each represent the states of any process from 1 to 2,000,000, then our two subjects become one.  The planets are entirely networked together.  But partial disjointedness is also possible: Maybe processes can represent the states of anyone within 1,000,000 of their own number (call this the Within a Million case).  Or maybe processes numbered 950,001 to 1,050,000 can be represented by any process from 1 to 2,000,000 but every process below 950,001 can only be represented by processes 1 to 1,050,000 and every process above 1,050,000 can only be represented by processes 950,001 to 2,000,000 (call this the Overlap case).

The Overlap case might be thought of as two discrete subjects with an overlapping part.  Subject A (1 to 1,050,000) and Subject B (950,001 to 2,000,000) each have their private experiences, but there are also some shared experiences (whenever processes 950,001 to 1,050,000 become sufficiently famous in the range constituting each subject).  Transitivity of Unity thus fails: Subject A experiences, say, a taste of a cookie (process 312,421 becoming famous across processes 1 - 1,050,000) and simultaneously a sound of a bell (process 1,000,020 becoming famous across processes 1 - 1,050,000); while Subject B experiences that same sound of a bell alongside the sight of an airplane (both of those processes being famous across processes 950,001 - 2,000,000).  Cookie and bell are unified in A.  Bell and airplane are unified in B.  But no subject experiences the cookie and airplane simultaneously.

In the Overlap case, discrete countability is arguably preserved, since it's plausible to say there are exactly two subjects of experience.  But it's more difficult to retain Discrete Countability in the Within a Million case.  There, if we want to count each distinct fame group as a separate subject, we will end up with a million different subjects: Subject 1 (1 to 1,000,001), Subject 2 (1 to 1,000,002), Subject 3 (1 to 1,000,003) ... Subject 1,000,001 (1 to 2,000,000), Subject 1,000,002 (2 to 2,000,000), ... Subject 1,999,999 (999,999 to 2,000,000), Subject 2,000,000 (1,000,000 - 2,000,000).  (There needn't be a middle subject with access to every process: Simply extend the case up to 3,000,000 processes.)  While we could say there would be two million discrete subjects in such an architecture, I see at least three infelicities:

First, person 1,000,002 might never be famous -- maybe even could never be famous, being just a low-level consumer whose destiny is to only to make others famous.  If so, Subject 1 and Subject 2 would always have, perhaps even necessarily would always have, exactly the same experiences in almost exactly the same physical substrate, despite being, supposedly, discrete subjects.  That is, at least, a bit of an odd result.

Second, it becomes too easy to multiply subjects.  You might have thought, based on the other cases, that a million processes is what it takes to generate a human subject, and that with two million processes you get either two human subjects or one large subject.  But now it seems that, simply by linking those two million processes together by a different principle (with about 1.5 times as many total connections), you can generate not just two but a full two million human subjects.  It turns out to be surprisingly cheap to create a plethora of discretely different subjective centers of experience.

Third, the model I've presented is simplified in a certain way: It assumes that there are two million discrete, countable processes that could potentially be famous or create fame in others by representing them.  But cognitive processes might not in fact be discrete and countable in this way.  They might be more like swirls and eddies in a turbulent stream, and every attempt to give them sharp boundaries and distinct labels might to some extent be only a simplified model of a messy continuum.  If so, then our two million discrete subjects would itself be a simplified model of a messy continuum of overlapping subjectivities.

The Within a Million case then, might be best conceptualized not as a case of one subject of experience, nor two, nor two million, but rather a case that defies any such simple numerical description, contra Discrete Countability.

#

This is abstract and far-fetched, of course.  But once we have stretched our minds in this way, it becomes, I think, easier to conceive of the possibility that some real cases (cognitively partly disunified mollusks, for example, or people with unusual conditions or brain structures, or future conscious computer systems) might defy transitivity of unity and discrete countability.

What would it be like to be such an entity / pair of entities / diffuse-bordered-uncountable-groupish thing?  Unsurprisingly, we might find such forms of consciousness difficult to imagine with our ordinary vertebrate concepts and philosophical tools derived from our particular psychology.

An Argument Against Every General Theory of Consciousness

As a philosophical expert on theories of consciousness, I try to keep abreast of the most promising recent theories. I also sometimes receive unsolicited emails from scholars who have developed a theory that they believe deserves attention. It's fun to see the latest cleverness, and it's my job to do so, but I always know in advance that I won't be convinced.

I'd like to hope that it's not just that I'm a dogmatic skeptic about general theories of consciousness. In "The Crazyist Metaphysics of Mind", I argue that our epistemic tools for evaluating general theories of consciousness are, for the foreseeable future, too flimsy for the task, since all evaluations of such theories must be grounded in some combination of dubious (typically question-begging) scientific theory, dubious commonsense judgment (shaped by our limited social and evolutionary history), and broad criteria of general theoretical virtue like simplicity or elegance (typically indecisive among theories that are live competitors).

Today, let me try another angle. Ultimately, it's a version of my question-beggingness complaint, but more specific.

Premise 1: There is no currently available decisive argument against panpsychism, the view that everything is conscious, even very simple things, like solitary hydrogen ions in deep space. Panpsychism is, of course, bizarrely contrary to common sense, but (as I also argue in The Crazyist Metaphysics of Mind) all well-developed general theories of consciousness will have some features that are bizarrely contrary to common sense, so although violation of common sense is a cost that creates an explanatory burden, it is not an insurmountable theory-defeater. Among prominent researchers who defend panpsychism or at least treat seriously a view in the neighborhood of panpsychism are Giulio Tononi, David Chalmers, Galen Strawson, and Philip Goff.

There are at least three reasons to take panpsychism seriously. (1.) If, as some have argued, consciousness is a fundamental feature of the world, or a property not reducible to other properties, it would be unsurprising if such a feature were approximately as widespread as other fundamental features such as mass and charge. (2.) Considering the complexity of our experience (e.g., our visual experience) and the plausibly similar complexity of the experience of other organisms with sophisticated sensory systems, one might find oneself on a slippery slope toward thinking that the least complex experience would be possessed by very simple entities indeed (see Chalmers 1996, p 293-7, for a nice exposition of this argument). (3.) Despite my qualms about Integrated Information Theory, there's an attractive theoretical elegance to the idea that consciousness arises from the integration of information, and thus that very simple systems that integrate just a tiny bit of information will correspondingly have just a tiny bit of consciousness.

Premise 2: There is no currently available decisive argument against theories of consciousness that require sophisticated self-representation of the sort that is likely to be absent from entities that lack theories of mind. On extreme versions of this view, even dogs and infants might not have conscious experience. (Again, highly contrary to common sense, but!) Among prominent researchers who have taken such a view seriously are Daniel Dennett and Peter Carruthers (though recently Carruthers has suggested that there might be no fact of the matter about the phenomenal consciousness, or not, of non-human animals).

There are at least three reasons to take seriously such a restrictive view of consciousness: (1.) If one wants to exit the slippery slope to panpsychism, one possibly attractive place to do so is at the gap between creatures who are capable of explicitly representing their own mental states and those that cannot do so. (2.) Consciousness, as was noted by Franz Brentano (and recently emphasized by David Rosenthal, Uriah Kriegel, and others), might plausibly always involve some sort of self-awareness of the fact that one is conscious -- apparently a moderately sophisticated self-representational capacity of some sort. (3.) There's a theoretical elegance to self-representational theories of consciousness. If consciousness doesn't just always arise when information is integrated in a system, an attractive explanation of what else is needed is some sort of sophisticated ability of a system to represent its own representational states.

Now you might understandably think that either panpsychism or a human-only views of consciousness is so extreme that we can be epistemically justified in confidently rejecting one or the other. If so, we can run the argument with weaker versions of Premise 1 and/or Premise 2:

Premise 1a (weaker): There is no currently available decisive argument against theories of consciousness that treat consciousness as very widespread, including perhaps in organisms with fairly small and simple brains, or in some near-future AI systems.

Premise 2a (weaker): There is no currently available decisive argument against theories of consciousness that treat consciousness as narrowly restricted to a class of fairly sophisticated entities, perhaps only mammals and birds and similar organisms capable of complex, flexible learning, and no AI systems in the foreseeable future.

Premise 3: All general theories of consciousness commit to the falsity of either Premise 1, Premise 2, or both (alternatively Premise 1a, Premise 2a, or both). If they do not so commit, then they aren't general theories of consciousness, though they may of course be perfectly fine narrow theories of consciousness, e.g., theories of consciousness as it happens to arise in human beings. (I've got a parallel argument against general theories of consciousness even as they apply just to human beings, based on considerations from Schwitzgebel 2011, ch. 6, but not today.)

Therefore, all general theories of consciousness commit to the falsity of some view against which there is no currently available decisive argument. They thereby commit beyond the evidence. They must either assume, or accept on only indecisive evidence, either the falsity of panpsychism, or the falsity of sophisticated self-representational views of consciousness, or both. In other words, they inevitably beg the question against, or at best indecisively argue against, some views we cannot yet justifiably reject.

Still, go ahead and build your theory of consciousness. You might even succeed in building the true theory of consciousness, if it isn't yet out there! Science and philosophy needs bold theoretical adventurers. But if a skeptic on the sidelines remains unconvinced, thinking that you have not convincingly dispatched some possible alternative approaches, the skeptic will probably be right.

ETA: In order to constitute an argument against a candidate theory, as opposed to merely an objection to such theories, perhaps I need to put some weight on the positive arguments in favor of views of consciousness that conflict with the theory being defended. Thanks to David Chalmers and Francois Kammerer on Facebook for pushing me on this point.

[image source]

A Model of Disunified Human Experience

It's a philosophical truism that human conscious experience is unified: If you're at a bar, hearing music, tasting beer, and feeling pleasantly relaxed, those experiences don't occur merely side by side. They are joined together into an integrated whole, an experience of music-with-beer-with-relaxation.

I'm not sure this truism is correct. As I suggested in an earlier post, experiential unity might be an artifact of introspection and memory: When we introspectively notice that we're experiencing music, beer, and relaxation all at once, we thereby bind those experiences into a whole. Likewise, when we remember such moments, we reconstruct them as unified. But it doesn't follow that those experiences, even if they all occurred simultaneously in you, were unified rather than transpiring separately. Experiences of music, beer, and relaxation might have all being going on inside of you, no more joined together than those experiences are joined with the similar experiences of your friend across the table. Simple co-occurrence doesn't entail experiential unity.

If this possibility is coherent, then introspection and memory can't establish that experience is always unified. At most, they show that introspected and remembered experiences present themselves as unified. But that leaves open the status of unintrospected, unremembered experiences. Unity becomes difficult to verify by standard phenomenological methods.

But the issue needn't be intractable. We just need to approach it less directly, for example by exploring what follows from a well-established theory of consciousness. If some well-motivated Theory X implies unity (or disunity), that would provide reason to accept its conclusion.

I'll now present a candidate Theory X. I'm not suggesting that this is the right theory of consciousness! For one thing, it's simplistic. I'm sure the mind is much more complicated than I'm about to say. I offer this theory only as a proof of concept. There could be a theory of consciousness with massive disunity as an implication.

This theory combines Global Workspace Theory and Recurrent Processing Theory. According to this hybrid, Global Workspace Theory governs attended experiences -- those targeted by introspection or reconstructed in memory -- while Recurrent Processing Theory governs unattended experiences.

The mind, on this picture, is composed of many separate "modules" that work mostly independently, connected by a workspace where a small amount of attended information is shared globally. There's a visual module, an auditory module, modules for motor activity, episodic memory, and so on. When we attend to something -- say, the taste of beer -- the information from the relevant module is broadcast into the Global Workspace, where it can be accessed by and influence processes in all the other modules. When unattended, the information stays local.

Here's one illustration of this type of architecture:

[the Global Workspace; source]

Orthodox Global Workspace Theory holds that only what is broadcast into the workspace is conscious. Theory X alters that assumption. Many people hold that conscious experience vastly outruns attention. Many people hold, that is, that you can experience the hum of traffic in the background when you're not attending to it, and the feeling of your feet in your shoes, and the leftover taste of coffee in your mouth, etc. -- all in a peripheral way, simultaneously, when your focus is elsewhere. Theory X, drawing on Recurrent Processing Theory, holds that such processes are conscious whenever there's enough cognitive activity of the right sort (recurrent processing, for example) in the modules, even without global broadcast.

The picture, then, is this: We have multiple sensory (and other) experiences all running simultaneously, each with enough cognitive processing to be conscious, but few of which are selected for global availability through attention.

Is there reason to think these modular processes are unified with one another? I see no reason to think so, if they're genuinely modular -- that is, if their processing stays local, exerting little influence elsewhere. The taste-of-beer processing stays in the tasting module. The sound-of-music processing stays in the auditory module. No link up. No straightforward causal, functional, or physiological basis for a unified experience of beer-with-music rather than, separately, an experience of beer and an experience of music.

When we introspect the beer and music simultaneously, we pull both into the Global Workspace, and there they unify. We might then mistakenly think they were unified all along, but that's an illusion of introspection. It's an example of the "refrigerator light error", the error of thinking that the light is always on because it's always on when you open the door to check.

On this model, disunity is the normal human condition. Our experiences are fragmented, except when we pull them together through attention. We just don't realize that fact because, so to speak, we only attend to what we attend to.

Two caveats:

First, this is probably not the right model of consciousness. But I don't think it's unreasonable to wonder if the correct model is similar enough to have the same implications. If so, we can't simply accept the unity of consciousness as a given.

Second, the recurrent peripheral, modular processes that don't make it into the workspace might not be determinately conscious. They might be only borderline conscious, in the indeterminate middle between consciousness and nonconsciousness, like a color can be indeterminately between green and not-green. This opens a third possibility, alongside unity and disunity: unity among the determinately conscious experiences with a hazy penumbra of indeterminate experiences that remain disunified. (There are further possibilities beyond these three; but save them for another day.)

Three Epistemic Problems for Any Universal Theory of Consciousness

By a universal theory of consciousness, I mean a theory that would apply not just to humans but to all non-human animals, all possible AI systems, and all possible forms of alien life. It would be lovely to have such a theory! But we're not at all close.

This is true sociologically: In a recent review article, Anil Seth and Tim Bayne list 22 major contenders for theories of consciousness.

It is also true epistemically. Three broad epistemic problems ensure that a wide range of alternatives will remain live for the foreseeable future.

First problem: Reliance on Introspection

We know that we are conscious through, presumably, some introspective process -- through turning our attention inward, so to speak, and noticing our experiences of pain, emotion, inner speech, visual imagery, auditory sensation, and so on. (What is introspection? See my SEP encyclopedia entry Introspection and my own pluralist account.)

Our reliance on introspection presents three methodological challenges for grounding a universal theory of consciousness:

(A.) Although introspection can reliably reveal whether we are currently experiencing an intense headache or a bright red shape near the center of our visual field, it's much less reliable about whether there's a constant welter of unattended experience or whether every experience comes with a subtle sense of oneself as an experiencing subject. The correct theory of consciousness depends in part on the answer to such introspectively tricky questions. Arguably, these questions need to be settled introspectively first, then a theory of consciousness constructed accordingly.

(B.) To the extent we do rely on introspection to ground theories of consciousness, we risk illegitimately presupposing the falsity of theories that hold that some conscious experiences are not introspectable. Global Workspace and Higher-Order theories of consciousness tend to suggest that conscious experiences will normally be available for introspective reporting. But that's less clear on, for example, Local Recurrence theories, and Integrated Information Theory suggests that much experience arises from simple, non-introspectable, informational integration.

(C.) The population of introspectors might be much narrower than the population of entities who are conscious, and the first group might be unrepresentative of the latter. Suppose that ordinary adult human introspectors eventually achieve consensus about the features and elicitors of conscious in them. While indeed some theories could thereby be rejected for failing to account for ordinary human adult consciousness, we're not thereby justified in universalizing any surviving theory -- not at least without substantial further argument. That experience plays out a certain way for us doesn't imply that that it plays out similarly for all conscious entities.

Might one attempt a theory of consciousness not grounded in introspection? Well, one could pretend. But in practice, introspective judgments always guide our thinking. Otherwise, why not claim that we never have visual experiences or that we constantly experience our blood pressure? To paraphrase William James: In theorizing about human consciousness, we rely on introspection first, last, and always. This centers the typical adult human and renders our grounds dubious where introspection is dubious.

Second problem: Causal Confounds

We humans are built in a particular way. We can't dismantle ourselves and systematically tweak one variable at a time to see what causes what. Instead, related things tend to hang together. Consider Global Workspace and Higher Order theories again: Processes in the Global Workspace might almost always be targeted by higher order representations and vice versa. The theories might then be difficult to empirically distinguish, especially if each theory has the tools and flexibility to explain away putative counterexamples.

If consciousness arises at a specific stage of processing, it might be difficult to rigorously separate that particular stage from its immediate precursors and consequences. If it instead emerges from a confluence of processes smeared across the brain and body over time, then causally separating essential from incidental features becomes even more difficult.

Third problem: The Narrow Evidence Base

Suppose -- very optimistically! -- that we figure out the mechanisms of consciousness in humans. Extrapolating to non-human cases will still present an intimidating array of epistemic difficulties.

For example, suppose we learn that in us, consciousness occurs when representations are available in the Global Workspace, as subserved by such-and-such neural processes. That still leaves open how, or whether, this generalizes to non-human cases. Humans have workspaces of a certain size, with a certain functionality. Might that be essential? Or would literally any shared workspace suffice, including the most minimal shared workspace we can construct in an ordinary computer? Human workspaces are embodied in a living animal with a metabolism, animal drives, and an evolutionary history. If these features are necessary for consciousness, then conclusions about biological consciousness would not carry over to AI systems.

In general, if we discover that in humans Feature X is necessary and sufficient for consciousness, humans will also have Features A, B, C, and D and lack Features E, F, G, and H. Thus, what we will really have discovered is that in entities with A, B, C, and D and not E, F, G, or H, Feature X is necessary and sufficient for consciousness. But what about entities without Feature B? Or entities with Feature E? In them, might X alone be insufficient? Or might X-prime be necessary instead?

The obstacles are formidable. If they can be overcome, that will be a very long-term project. I predict that new theories of consciousness will be added faster than old theories can be rejected, and we will discover over time that we were even further away from resolving these questions in 2025 than we thought we were.

[a portion of a table listing theories of consciousness, from Seth and Bayne 2022]