This is a preprint of a review published in The Journal of Consciousness Studies, 27 (1-2), 242-56. Please cite from the published version.
In Rethinking Consciousness, Graziano really is rethinking consciousness, and will undoubtedly inspire others to do the same. He certainly forced me into some enjoyable rethinking as well as painful struggling to work out just what his ‘Attention Schema Theory’ (AST) means for consciousness.
His first, less controversial, step is to stop asking where or how consciousness is generated and instead ask how and why we attribute consciousness to others and to ourselves. So he is not trying to solve the hard problem but attacking the ‘meta-problem of consciousness’ (Chalmers 2018), trying to understand ‘why people might mistakenly think that there is a hard problem to begin with’ (p 2). He is clearly taking an illusionist approach in the way he does so, although he does not readily admit to this.
Second, he makes a bold theoretical leap, suggesting that our metaphysical intuitions about our own consciousness – about being a ‘ghost in the machine’, with all manner of spooky powers – are derived from a specific internal model, the attention schema. This he likens to the more familiar body schema, which he and his colleagues have long been working on. The body schema is a model of our whole body, or indeed of a crab, an elephant or an octopus’s body, that keeps track of the body’s structure and position in order to control its movements. In engineering terms, this is essential. Any control system needs a model of what is to be controlled.
What Graziano adds is that the same is true of attention. Attention is a fundamental brain resource, and to control its deployment and focus we also need an attention schema. Both models are approximations, simplifications, and therefore make revealing mistakes. Phantom limbs are an example in the case of the body schema. Our bodily representation is so powerful that when people lose a limb, they may still feel its presence, and even suffer pain in a non-existent hand or leg. Graziano draws a close analogy between the body and attention schemata, “One is the ghost in the body and the other is the ghost in the head” (p 104).
What, then, is the connection between the attention schema and consciousness? This is a highly charged question given the long and confusing history of the relationship between attention and consciousness, with some theorists equating them while others describe complex causal or correlational relationships between them (Blackmore and Troscianko 2018). For Graziano, the relationship is clear, “Attention is something the brain does; consciousness is something the brain says it has.” (p 110). I’ll be finickity here and point out the mereological fallacy – it is not brains that say anything, but people who do. Nevertheless, the point is clear, when we say we have consciousness, we are wrong. This makes his theory a version of illusionism.
Why are we wrong? The cortex, Graziano explains, is an attention machine and the cortical attention schema is ‘a set of information about the process of attention itself’ (p 40). In doing so, it provides a cartoonish account of a complex process, and by simplifying what is going on describes an insubstantial essence with a location vaguely inside you, that grasps objects, sounds, thoughts and memories ‘and that restlessly moves about, searching, seizing some items and dropping others’ (p 42). The items it takes hold of become clear, real, and vividly present to you – they become your conscious experience and you can then understand, respond to, and act upon them. This seems clear but I want to dig deeper. Is there something it is like to be an attention schema? Or is the whole idea of subjective experience being ‘what it is like to be …’, part of the ‘cartoonish account’ too?
Part of the strength – and fun – of this theory is that it can explain why you seem to have an amorphous ‘power of consciousness’, an ethereal mental essence that can reach out into the world. It shows why you struggle to say much about your own subjective awareness other than the things you are aware of. This leads Graziano to some insightful comments on why we believe weird things. I especially enjoyed his take on the extramission theory of vision. This is the ever-popular idea that vision involves a beam emerging from the eyes to light upon the objects seen. This seems to be the default belief of young children, although most educated people soon learn that it is not true. To explore this effect, he describes a clever experiment in which volunteers looked at upright tubes, in various sizes and proportions, and had to judge the angle at which they would fall over. On each trial a face in profile was seen looking at the tube. The odd effect they found was as though an ‘eye beam’ was coming from the face, helping to prop up the tube if it was tipping towards them or to help it fall over if it was tipping away. When the face was seen to be blindfolded there was no such effect. Most of the participants correctly thought that vision involved light coming into the eyes, but still showed the effect. In this way Graziano reveals how we come to believe in eye-beams and in an invisible, gentle, mind-force emanating from faces.
A potential strength of the theory is that it builds on, rather than rejects, previous theories, including higher-order thought, integrated information, and especially global workspace theory (GWT). Graziano calls the workspace, ‘the gossip central of the brain’, that achieves ‘fame in the brain’ (Dennett 2001). ‘Information that has entered the global workspace has also entered consciousness’ (p 105). Yet, he says, global workspace theory is incomplete because it does not explain why or how such information becomes conscious. ‘The attention schema theory offers a way to complete the picture’ (p 106).
This is a great claim but hits a serious problem because GWT is fundamentally ambiguous and I am not sure which picture Graziano thinks he is completing. One interpretation of GWT relies on a magical transformation: something enters the GW “and then it ‘becomes conscious’ or ‘enters consciousness’” with no explanation of what this means (Blackmore and Troscianko 2018 p 116). People who see it this way, and they may be in the majority, are like Dennett’s ‘Cartesian materialists’ who believe in a ‘crucial finish line or boundary somewhere in the brain’ where what happens ‘is what you are conscious of’ (Dennett 1991 p 107). The harder, and less intuitive, interpretation is to say that once something is in the GW it is widely broadcast, providing access to further processing, and to outputs such as action and speech. There is no special magic – such access is all consciousness is. This is what Dennett means by ‘fame in the brain’ or ‘cerebral celebrity’ (Dennett 2001). Fame is not something additional to being widely known; consciousness is not something additional to global availability. It is an illusion to think it is.
So which picture does Graziano claim to be completing? In some places he implies the latter, taking an illusionist stance and specifically mentioning ‘fame in the brain’. Yet elsewhere he falls back on magic, criticising GWT because ‘it doesn’t offer any specific explanation for why information, having arrived in the global workspace, gets the property of conscious experience attached to it.’ (p 105). I find this utterly confusing because, by his own theory, or any illusionist theory, consciousness is not the kind of thing that can be ‘attached’ to anything.
If we ignore this confusion, does linking AST to GWT still have some advantages? Potentially AST can build on rather than compete with GWT, but I fear that even this may lead to trouble. GWT has the advantage of simplicity in one respect; it claims that the contents of the GW are the contents of consciousness. Yet Graziano must lose this simple equation because his attention schema models only partially, and in simplified form, the contents of the GW. He likens the GW to the scary, legendary sea-monster, the Kraken, and the AS to the real giant squid upon which the myth is probably based.
Another problem is that tying AST to the GW narrows its scope and flexibility because the attention schema can model only attentional processes going on within the GW. This wouldn’t matter if everything the brain allocates its attention to is automatically in the GW but this might not be true.
I think of oddities like the unconscious driving phenomenon, when you arrive at your destination with a full memory of the interesting conversation you had with your passenger, or the gripping news on the radio, yet remember nothing about driving. What was in the GW? Apparently not the driving because the contents of the GW are conscious. Yet some parts of the brain must have been paying close attention to the road or you might have jumped a red light, missed a turning, or run over an old lady. All this must surely need an attention schema too. It seems to me that Graziano must either say that there were two (or more) conscious streams and one was inaccessible or forgotten, or explain why this other attention schema is not reported as having been conscious. The same applies to many things we do while concentrating on something else.
Then there are unusual states of consciousness reached in meditation, psychedelic drugs, or mystical experiences, when thoughts fall silent leaving only spaciousness, or when self disappears into oneness. In such states there may be no global workspace at all, and no self model, yet people often say that these were the most intensely conscious experiences in their life. As a long-term meditator, I am enjoying exploring various altered states while asking myself what kind of attention schema might be entailed and I believe the theory might help us understand such states, but this seems harder if the theory is tied to GWT.
Coming back to earth, AST helps with questions about animal and machine consciousness. These are tricky because we have no way of asking a frog or a laptop whether it is conscious, of understanding its answers if it gave any, or of independently detecting its consciousness. Graziano traces the long evolution of the cortex, or ‘attention machine’, from mammal-like reptiles 300 million years ago to the present. If any of these build attention schemata, then they will have some form of consciousness.
The same applies to machines. If you build a machine according to the theory then “The machine will think it has consciousness, claim to have consciousness, and talk about its consciousness, because you will have built the construct of consciousness into it.” (p 119). This Graziano welcomes. He’d rather live in a world full of machines that can attribute consciousness to themselves and to him. Indeed, he thinks that building such machines is the most useful way ‘to reduce the technological risks looming in our future’ (p 137).
And leaving earth? In a final chapter, Graziano expresses little optimism for the prospect of mind-uploading based on the human connectome. To have a sense of self, to still be ‘me’, we need bodies, so he explores the prospect of building simulated brains and super-realistic virtual bodies embedded in convincing virtual worlds. Many pitfalls await, but he imagines a future in which young people remain on earth in a kind of ‘larval stage’ before moving on to the cloud world where life really begins; a world in which minds are grown, nurtured, saved and even merged with other minds, dispersing through space and exploring the galaxy.
I have really enjoyed Graziano’s ideas but I wish he had more warmly welcomed the fact that AST is a form of illusionism. Instead, having briefly admitted as much, he distances himself from the idea because “calling consciousness an illusion is the kiss of death for a theory (p 97). Admittedly illusionism is hard to understand and difficult to accept. But is that a reason to disguise it? If we think consciousness is an illusion, as I do (Blackmore 2018), we should be bold in saying so and then help people to understand it. AST not only claims that people are wrong about their own experience but explains how and why they are wrong. This is what is illusionism and I am sorry Graziano was not more positive about it.
Does this matter? I believe so, because while people (and I include many consciousness researchers here) go on thinking that we need to explain how consciousness arises from, or is made by, or emerges from, the brain, we are stumped. Because it doesn’t. We just imagine that it does.
AST may not solve all the problems Graziano would like to solve but it is an important advance and, despite my reservations, I am much looking forward to seeing where it takes us.
Blackmore, S. (2016). Delusions of consciousness. Journal of Consciousness Studies, 23(11-12), 52-64.
Blackmore, S., & Troscianko, E. T. (2018). Consciousness: an introduction. Routledge.
Chalmers, D. J. (2018). The meta-problem of consciousness. Journal of Consciousness Studies, 25(9-10), 6-61.
Dennett, D. (1991) Consciousness Explained. Little, Brown & Co.
Dennett, D. (2001). Are we explaining consciousness yet? Cognition, 79(1-2), 221-237.