Dan Dennett – Self as a Centre of Narrative Gravity

Image result for self as centre of gravity

Dan Dennett’s 1992 article entitled “The Self as a Center of Narrative Gravity” (available online here) is one of his earlier attempts to argue that the self isn’t real; specifically, that it is a fiction created by human beings who don’t know any better. The route he takes to get there is a somewhat jumbled, confused batch of mixed metaphors, semi-relevant thought experiments, and false implications. In this article, I will try to unravel the tangle Dennett gets himself into and in doing so, resist his conclusion.

Dennett’s Article

Centres of Gravity

Dennett starts by describing the physical concept of a centre of gravity. It is not, he correctly points out, “an atom or a subatomic particle or any other physical item in the world.” (p.1) Lacking all physical properties except for a spatio-temporal location, it is “an abstractum… a fictional object” (p.2) Likewise, Dennett claims; “A self is also an abstract object, a theorist’s fiction.” (p.2)

Fictional Characters

Dennett then offers another analogy, this time comparing the self to a fictional character in literature. Of course, fictional characters, by definition, don’t really exist. Dennett illustrates this with the principle of bivalence, which states that any valid proposition must be either true or false; i.e. it can’t be both true and false, nor can it be neither true nor false. Now, consider the question: Did Sherlock Holmes have a mole on his left shoulder blade? Arthur Conan Doyle never said anything about this and we can’t infer its truth or falsity from the books. So, what are we to make of this? Well, if Holmes was a real person, even if, for whatever reason, we could never discover the truth, we could still be absolutely certain that there was a truth to the matter. But since Holmes was a fictional character, there just is no truth to be found here. It’s neither true nor false. The principle of bivalence doesn’t apply. If you find yourself seriously wondering about Holmes’ left shoulder blade, you have failed to understand what fictional characters are. Centres of gravity, as fictional objects, are the same; “If you scratch your head and say, “I wonder if maybe centers of gravity are really neutrinos!” you have misunderstood the theoretical status of a center of gravity.” (p.3)

The Mind is a Dumb Machine

Next, in order to make the link between fictional characters and selves explicit, and rule out the objection that “fictional selves [are] dependent for their very creation on the existence of real selves” (p.3), Dennett takes us on a thought experiment. Imagine an AI that has been designed to write novels. Even though it can write perfectly consistent and readable novels, Dennett is clear that it isn’t conscious. “It is a dumb machine, but it does have the power to write a passable novel.” (p.3) So, fictional characters can, at least in theory, come about from a self-less, non-conscious object.

If our AI then starts incorporating its own real-life experiences into the (fictional) main character’s life, what we have is a selfless, non-conscious object creating something we can recognise as a fictional, narrative self, even though it has no idea what it is doing. “We can still maintain that the robot’s brain, the robot’s computer, really knows nothing about the world; it’s not a self. It’s just a clanky computer. It doesn’t know what it’s doing. It doesn’t even know that it’s creating a fictional character.” (p.4) Dennett now makes the bold claim that this is exactly what is happening with us all the time; “The same is just as true of your brain; it doesn’t know what it’s doing either.” (p.4)

The Modular Theory of Mind

Dennett next moves on to criticise the idea that our minds are a unified whole. Citing neuroscientist Michael Gazzaniga, he argues that the human mind is “not beautifully unified, but rather a problematically yoked-together bundle of partly autonomous systems.” (p.6)[1] Gazzaniga’s research on split-brain subjects – people whose corpus callosum has been severed, preventing the two hemispheres of the brain communicating with each other – certainly offers compelling proof that the two hemispheres are autonomous and capable of operating independently of each other. (In The Consciousness Instinct Gazzaniga reports how one subject even had to sit on his left hand to stop his right hemisphere from interfering with the task he was trying to complete with his right hand). Ordinarily, we don’t experience the disunity and struggle between these different “modules” but nevertheless, it is what is really happening. Dennett offers further support for the truly fragmented nature of the human mind by referring to the well-documented instances of people suffering from Multiple Personality Disorder.

David Hume

The article concludes with Dennett quoting the modern sceptic, David Hume, who observed that:

For my part, when I enter most intimately into what I call myself, I always stumble on some particular perception or other, of heat or cold, light or shade, love or hatred, pain or pleasure. I never can catch myself at any time without a perception, and never can observe anything but the perception…. If anyone, upon serious and unprejudiced reflection, thinks he has a different notion of himself, I must confess I can reason no longer with him. All I can allow him is, that he may be in the right as well as I, and that we are essentially different in this particular. He may, perhaps, perceive something simple and continued, which he calls himself; though I am certain there is no such principle in me.


Centres of Gravity

Let’s start with Dennett’s centre of gravity. While we can all agree that the centre of gravity of an object is not equivalent to an actual physical particle or collection of particles, Dennett is playing pretty fast and loose with the word ‘abstract’ when he names it so. ‘Abstract’ means existing only in the mind; i.e. lacking physical or concrete existence. True abstract concepts are things like love, beauty, number, etc. None of these things exist in the physical world. You can’t point to love or the number 2 anywhere, even if you’re standing in front of two people in love with each other. On the other hand, we can easily point to the centre of gravity of a bat, for example, even as we acknowledge that the physical point we have identified is not identical with the atoms that are under our finger. In fact, centres of gravity can only be physical! Dennett admits a centre of gravity has a “spatio-temporal location” while at the same time claiming it is “a purely abstract object.” This is surely nonsense on stilts.

So, a centre of gravity is not really an “abstractum”. What about the subsequent claim that it is also a fiction? This is again a slippery use of words. The word ‘fiction’ is heavily loaded, something Dennett is surely aware of, but, with his broad brush, he fails to make explicit any of the subtleties at play. In ordinary speech, to call something a fiction is to say it is wholly invented by its author and therefore not real. The paradigmatic example of fiction is story-telling. A fictional story is a story that never happened and a fictional character is a ‘person’ that never existed. Importantly, however, this is not a technical definition. It isn’t equipped to carry the weight of any scientific or philosophical claims, but this is exactly what Dennett demands of it. He co-opts the word to mean something that doesn’t physically exist; i.e. as a collection of particles, and then also picks up, for free as it were, the notion of unreality; it doesn’t physically exist, therefore it isn’t real.

We can easily see how Dennett’s use of the word is illegitimate by considering another common, everyday phenomenon; thoughts. Are thoughts, which by definition, lack physical reality, fictions? If I think of the restaurant I visited yesterday, is that thought a fiction? The restaurant isn’t a fiction – it really exists. My thought about the restaurant also really existed – if it didn’t, we would be forced into the absurd conclusion that a non-occurrence occurred when I was thinking about it.

In reality, the question just doesn’t make sense. While the content of a thought can be fictional or factual, the thought itself is neither. You can make a fictional character but you can’t have a fictional thought. Is language fiction? What about love? Hopefully it’s clear that the word ‘fiction’ is completely inadequate when it comes to describing such phenomena. Returning to centres of gravity, even if we grant Dennett’s initial claim that they are pure abstracta, centres of gravity clearly aren’t fictional the way characters in a novel are. On the contrary, centres of gravity describe very real features of very concrete objects.

The Self

Dennett’s final move in this initial salvo is to make the analogy from centres of gravity to the self, which he claims is both “an abstract object” and “a theorist’s fiction.” (p.2) Given that I’ve denied that centres of gravity are either abstract or fictional, this analogy is basically irrelevant. However, we still need to investigate the self on its own merits.

In order to properly review Dennett’s claims about the self, we need to clarify exactly what we mean when we use the word. It seems to me that there are actually two senses of the word ‘self’ we have to distinguish between; the self I am, and the self I, or other people, reflect on. The former is our primary conscious awareness of ourselves as subjective experiencers. The latter is a second order construction we make when we think about the self that we are.

The self I am (I will use Dan Zahavi’s term, “minimal self”, to describe this) is essentially an awareness that “I” am experiencing something. It is a phenomenological fact that is part of what it means to be a subjective consciousness. If you claim not to have this awareness, you aren’t a conscious human being. If you claim you have it, but think it is an illusion, you haven’t understood what we are talking about. The curious thing about subjective experiences is that simply having them is sufficient and necessary for them to be real. All conscious experiences can be divided into two parts; the experience itself and the content of that experience. The former must be true and real. If it could be an illusion, you would be capable of experiencing a non-event. On the other hand, the content of that experience can be an illusion. You may, for example, have the conscious experience of being an immaterial soul in a physical body. What is illusory here is not the conscious experience qua experience, but your interpretation of the experience, which leads to the belief that you are an immaterial soul. An experience, by virtue of being an experience, is absolutely true. This is the minimal self.

The ‘reflected self’, on the other hand, is the result of an intellectual exercise in which we reflect on those subjective experiences and bring them together into a coherent totality. This presents us with two immediate consequences. First, falsely reporting (and even fabricating) experiences becomes both possible and tempting. Second, unlike the minimal self, the reflected self isn’t a private phenomenon. Other people, not reflecting on our subjective experiences, but reflecting on our overt behaviour, form their own opinions about who we are; in effect, re-making our ‘selves’, but from their perspective.

All of the above will necessarily inform our discussion as we continue to look at Dennett’s claims. It is interesting to note that Dennett himself doesn’t make this distinction because the only ‘self’ he recognises is the reflected-on self. With all of that, let’s see if the self is, in fact, “an abstract object” and “a theorist’s fiction.”

First, the minimal self. It’s tempting to call this an abstract object simply because it can’t be experienced through any of the five senses. The minimal self is intangible, but it is by definition, never an object; rather, it is that by which we posit objects. An abstractum is an object or idea that exists only in the experiencing mind. The minimal self, on the other hand, is that mind. Now, of course, we can imagine ourselves in our mind, but the self that is doing the imagining obviously can’t also be the object imagined. You can never get behind yourself to see yourself. No matter how far back you step, your minimal self, like the end of a rainbow, is always one step further back. It’s like looking at yourself in a mirror. Even though (most) mirrors faithfully reflect your body, you would never make the mistake of thinking that you are your reflection.

If the minimal self isn’t an abstraction, how about a fiction? Remember, the criteria for being fictional isn’t lacking physical existence, it is never existing at all. Clearly, the minimal self, as the experiencer you are, has experienced, is experiencing now, will experience in the future, and has been/is/will be aware of all of this experiencing in the first person… ergo the minimal self isergo it is not a fiction.

Let’s turn now to the reflected self. It’s just as tempting to think of this as an abstract object. Not only is it unavailable to the physical senses, but it’s also a second-order creation established by the minimal self. However, no matter how I twisted and turned this over in my mind something just didn’t quite sit right. Abstract objects (courage, number, etc.) are all not just non-physical, they’re also universal concepts that admit of specific instances. Courage in the abstract is an idealised, non-instantiated definition which is sufficiently broad to encompass any number of specific, concrete, courageous feelings and actions. The same also holds for number. ‘Two’ is an abstract concept, but the two books on my desk are very concrete.

If we apply this same criterion to the reflected self, we see that it doesn’t have this abstract/concrete or universal/particular relationship. The reflected self isn’t a universal concept we see instantiated in specific instances. On the contrary, the reflected self is, from the beginning, by definition, highly specific, being the reflected self of a single person. It’s true that it isn’t physical, but neither is it abstract.

Is it a fiction then? Let me answer this question with another question; if I tell a third party about that time you hit the winning runs in that game of cricket (and you actually did), is that account fictitious? What about where I tell the same story in the event you never hit the winning runs; i.e. I lie? Is that fictitious? Hopefully, you answered the first is not fictitious but the second is. Generalising to the reflected self then, tells us that the reflected self is only a fiction when the account of the self I (or someone else) give is untrue. An honest, unbiased reflected self, on the other hand, is not a fiction.

Fictional Characters

Dennett is right that fictional characters don’t adhere to the principle of bivalence; the question of whether or not Sherlock Holmes had a mole on his left shoulder blade is neither true nor false because Sherlock Holmes was never a real person. Dennett then claims that the principle of bivalence doesn’t hold for centres of gravity either because to ask whether centres of gravity are really neutrinos is to have misunderstood what centres of gravity are. This is pretty sloppy reasoning. The principle of bivalence says that a proposition must be either true or false, and the proposition above concerning centres of gravity, by Dennett’s own reasoning, is clearly false. Therefore bivalence holds. Interestingly, this supports my earlier claim that centres of gravity are not fictional.

The Mind is a Dumb Machine

There is nothing wrong with Dennett’s thought experiment here. A non-conscious computer could, at least theoretically, compose a perfectly consistent and readable novel, thereby producing fictional characters. The problem, as I’ve already argued, is that this has very little to do with the self because the minimal self and the (honestly recalled) reflected self aren’t fictions.

The really disturbing claim in this section comes in the very last sentence of my summary; namely, the claim that the human brain is no different from the non-conscious AI, the “dumb machine” that doesn’t know what it’s doing when it creates the fictional character we are supposed to believe is the equivalent of the selves we all think we are. It’s hard to know what to make of such an outrageous claim. Dennett is quite literally suggesting that consciousness is an illusion. Even though we think we are conscious (which, at its core, means being capable of, and aware of oneself, thinking about things), we aren’t. However, as I intimated in an earlier section, the fact that I am self-aware; i.e. that I am aware of myself and what I am doing, is self-proving. You can’t be mistaken about being self-aware and conscious because ‘being mistaken’ is a conscious state! Non-conscious objects; trees, books, Dennett’s “clanky computer”, etc., can’t be mistaken about their conscious states precisely because they aren’t conscious.

I can’t really think of anything else to say on the subject except that it’s a little ironic that Dennett is using his gift of consciousness to pretend he doesn’t have it. Food for thought though: could a race of non-conscious beings ever create a civilisation and have debates about the self? (Three things to add: 1) Dennett thinks yes (and his ‘proof’ is himself), 2) I think no (and my proof is every human being except Dennett), and 3) the answer doesn’t change any of my prior conclusions about Dennett’s article)

The Modular Theory of Mind

Dennett’s interpretation of the modular theory of mind suggests that rather than being a unified self, the human mind is really a “problematically yoked-together bundle of partly autonomous systems.” Split-brain patients and MPD sufferers support this thesis. The error here seems to me to be the false dichotomy Dennett sets up; the self is either perfectly whole and unified or there is no self. But couldn’t the self just be this “problematically yoked-together bundle of partly autonomous systems.” Does it matter how the sense of a coherent self is generated or is the fact that it is generated enough?

We can imagine any number of counter-examples to refute Dennett’s claim here. A thought is really just the firing of a bunch of neurons, therefore that house you were just thinking of is an illusion. A painting is really just little dots on a canvas, therefore the smiling woman you see is an illusion. That house you were thinking of is the firing of neurons, the smiling woman is those dots on a canvas, the self is the way the brain somehow yokes together disparate autonomous systems into a meaningful entity. Pointing out the micro-scale activity of the brain doesn’t eliminate, or render any less real, its macro-scale phenomenology.

Likewise, given my postulate of the minimal and the reflected selves, there is absolutely nothing preventing multiple selves from arising from the same brain at different times. On the contrary, the fact that there are credible accounts from people for whom this is the case suggests this is an idea we may want to embrace. It certainly doesn’t suggest the self is a fiction.

David Hume

The quote from David Hume is striking in that, rather than refuting the self, it seems to be saying almost exactly the same thing I said earlier about the minimal self. Recall how Hume says that no matter how closely he introspects, he can never catch himself anywhere. That is precisely the case with the minimal self (if you’ll forgive me quoting myself); “You can never get behind yourself to see yourself.” Like the absent-minded professor looking for the glasses he was wearing all along, Hume was looking for the self he was being all along.


Dennett’s naturalistic materialism prevents him from understanding what we all know intuitively. Like any good scientist, Dennett abstracts out from the first person perspective and approaches the problem of the self as if it could be solved from the third person perspective. Phenomenologists or anthropologists, he thinks, see people (“complicated things”) moving about in the world and, realising that it would be “theoretically perspicuous to organize the interpretation around a central abstraction” (p.2), postulate a fictitious self which lets them understand and interpret these “complicated things”. It is only after positing ‘selves’ for these human-shaped objects around us that we then turn inwards and “posit selves for ourselves as well.” (p.2)

Surely, Dennett has this around the wrong way. Nobody, prior to materialists like Dennett, has ever inferred their ‘selfhood’ only after inferring selves in the “complicated things” they saw around them. Dennett looks for the self ‘out there’ in the physical world, even trying to understand his own self from a third person perspective, and failing to find one concludes it doesn’t exist. It must be a fiction we invented merely to help us understand these complicated machines that claim to be conscious but clearly can’t be (because there is no consciousness ‘out there’ either). He fails to find the self, not because it doesn’t exist, but because he’s looking for it in the wrong place with the wrong tools.

[1] Gazzaniga’s most recent book, The Consciousness Instinct, is an excellent account of this.


5 thoughts on “Dan Dennett – Self as a Centre of Narrative Gravity

  1. Thanks for the critique, altho I find it quite off in some cases. For example, the center of mass is an abstraction, but maybe a different one than beauty (maybe beauty is just more abstract than a center of mass.) Still I could say that you can point to beauty when you see it. Maybe you can’t point to justice that easily. Regardless, he uses the center of mass as an analogy, but no analogy is perfect. I could still think we can defend the self as a very very abstract abstraction 🙂

    Liked by 1 person

    • Great comment. It seems to me that instead of being a critique Nathan should have instead written his article more as a clarification. It seem to me the difference between the two is for example in their understanding of abstraction and fiction which are fluid depending on the understanding of those words, which will inevitably change over time and by reader. Another challenge is that the slightly different interpretations of those words can result in a somewhat different conclusion…. The real answer seems to me to be in the grey area between the two points of view.

      Liked by 1 person

      • Hi T, thanks for the comment (on the comment). Allow me to briefly respond:

        Understandings are fluid. I agree with the sentiment here; language isn’t black and white. I just don’t think this is the problem. I mean I doubt Dennett and I really disagree about what the words ‘abstraction’ and ‘fiction’ mean. As I said in my response to Sasakitanjiro, Dennett’s whole argument turns on us being clear about these terms because he eventually wants to say that the self is an illusion (i.e. abstract and fictional). If we compromise these words before he gets there, we also compromise his final argument (a point you also note).
        You can disagree with me on whether centres of gravity are abstract or fictional (as Sasakitanjiro did), but I’m not sure you can write us both off by saying that understandings “inevitably change over time and by reader”, as if we’re both just being dogmatic and/or pedantic.


    • Sasakitanjiro: Thanks for reading the article and the taking the time to comment. I didn’t respond to you originally, but I wish I had because it turned out to be quite a fruitful exercise.

      First, we can’t point to beauty. We can point to beautiful things. This is a basic, but important, distinction.

      Second, is the centre of gravity a less abstract abstraction than beauty? This is a possible line to take; however, it still runs into the bigger problem you didn’t mention in your response; i.e. the centre of gravity having a “spatio-temporal location.” You can opt to bite the bullet on this (like Dennett), but for me it is a bridge too far to call something obviously concrete (I can’t think of a better definition for ‘concrete’ than ‘having a spatio-temporal location’) an “abstractum”. (Perhaps squares are circles as well, just less circular ones?) Moreover, this interpretative generosity doesn’t help Dennett because he needs abstraction to mean what we all know it means; i.e. lacking physical existence. Why? Because he ultimately wants to say that the self lacks physical existence. If we equivocate on the word ‘abstract’ in the argument, we equivocate on the word in the conclusion as well.

      Third, you make the point that the centre of gravity is used by Dennett as an analogy, and no analogy is perfect. Dennett uses the centre of gravity as an analogy because he believes it “has some properties in common with selves”; namely, being abstract and fictional. I didn’t invoke any irrelevant properties in my analysis, so I can’t see how this applies to my article. But this is a moot point anyway because I argued, independently of the analogy, that selves are neither abstract nor fictional.


  2. Pingback: L’io è una costruzione narrativa – Idee in Rete

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s