Multiagent Systems: Plato’s Chariot, AI, and Self in Information Philosophy

We will liken the soul to the composite nature of a pair of winged horses and a charioteer. […] the charioteer of the human soul drives a pair, one of the horses is noble and of noble breed, but the other quite the opposite in breed and character. Therefore in our case the driving is necessarily difficult and troublesome.

– Phaedrus 246a – 254e

Luciano Floridi introduces Plato’s chariot as a  problem in engineering, a technological problem to be analyzed not in the older modes of a phenomenological or descriptive approaches to the self but rather in the sense of postmodern constructionist or design-oriented approaches.1 Michael Wooldridge gives us a good working definition of a multiagent systems:

Multiagent systems are systems composed of multiple interacting computing elements, known as agents. Agents are computer systems with two important capabilities. First, they are at least to some extent capable of autonomous action – of deciding for themselves what they need to do in order to satisfy their design objectives. Second, they are capable of interacting with other agents – not simply by exchanging data, but by engaging in analogues of the kind of social activity that we all engage in every day of our lives: cooperation, coordination, negotiation, and the like.2

For Floridi many of the challenges in the engineering of these multiagent systems (MAS) couched in the new metaphors of AI research and engineering theory and practices stem from older forms philosophical speculation on the self. He cites such issues and concerns as communication, coherence, rationality, successful interaction with the environment, coordination and collaboration with other agents among others as crossover problems of classic philosophy. As he tells us their is another component beyond the charioteer and the two horses that Plato mentions that needs to be included in any notion of a multiagent conception of the self, and that is the ‘chariot’ itself. For it is the chariot that “guarantees the unity and coordination of the system, thus allowing the self to be, persist and act as a single, coherent, and continuous entity in different places, at different times, and through a variety of experiences” (Floridi, 5).

He informs us that up to Descartes time philosophers were concerned with the epistemological aspects of the self, but from the time of the rationalists the ontological aspects of self as a unified, persistent, and harboring a unique identity over time (diachronic dimension) came into play. In modern times two forms of analysis would emerge out of the Platonic egology and his Christian inheritors during the Age of Reason and after: the first, termed diachronic egology, concerned an ontology of personal identity, concentrating on the problems arising from the identification of a self through time or possible worlds, progressively moving towards metaphysics; the second, synchronic egology, deals with the individuation of a self through temporal dimensions or in a possible world, thus placing itself at the heart of the what would come to be known as the philosophy of mind(Floridi, 6).

He explains that the diachronic egology has certain flaws and incompatibilities to be considered within the context of an informational theory of the self. Historically this form of analysis of the self has taken one of two paths. Endurantism, a  view of material objects as persisting three-dimensional individuals wholly present at every moment of their existence. And, perdurantism or four dimensionalism which maintains that an object is a series of temporal parts or stages. For Floridi the main difficulty with either of these is that it tries to maintain an abstract free context in its description of these temporal identities which is impossible to maintain and leads to nonsense. He gives several examples but I’ll leave that to the reader to follow up on (i.e., link to original paper below).

As for synchronic egology Floridi mentions both the Lockean (sovereign self), and what he terms the Narrative approach to selfhood. In this he offered a passage from Proust’s Overture of which I’ll only recount a singular passage where he tells us that “none of us can be said to constitute a material whole, which is identical for everyone, and need only be turned up like a page in an account-book or the record of a will; our social personality is created by the thoughts of other people”. The point is that identification comes through informational contexts in which others become the key elements in the analysis.

As Floridi maintains “in both cases, individuation – the characterization or constitution of the self – is achieved through forms of information processing: consciousness and memory are dynamic states of information, but so is any kind of personal or social narrative” (Floridi, 9). One might even term this the cyborg self a sort of distributed jumble of prostheses, genes, hybrids, hierarchies, and parallel processors with information as the algorithmic control system guiding all the basic functions. Floridi quotes form Hume, who in a different time and framework understood the basis of information theory in describing his inability to find a unity to the self: “there are two principles, which I cannot render consistent; nor is it my power to renounce either of them, viz. that all our distinct perceptions are distinct existences and that the mind never perceives any real connection among distinct existences”:

…and that the mind never perceives any real connexion among distinct existences. Did our perceptions either inhere in something simple and individual, or did the mind perceive some real connexion among them, there would be no difficulty in the case. For my part, I must plead the privilege of a sceptic, and confess, that this difficulty is too hard for my understanding.3

As Floridi states it “if the self is made of information (perceptions or narratives, or any other informational items one may privilege), then a serious challenge is to explain how that information is kept together as a whole, coherent, sufficiently permanent unit” (Floridi, 10). Maybe, as my friend R. Scott Bakker, keeps saying, we should just quit trying. Maybe the self is a temporary agent produced by the brain in its ongoing activities, along with the memory sequences attached to this temporary locus of activity as an artificial agent to handle specific evolutionary intermediation between brain and environment and nothing more (i.e., after the specific activity subsides so does the self with it). Our sense of continuity, of individuation and change might in the end all be distortions of an information processing system to vast and inaccessible to either internal or external reflection that we have yet to comprehend its complexity.

As Floridi tries to suggest if in this notion of synchronic ecology with the self as narrative –  what prevents the narrative from being a completely random, incoherent and disjointed selection of miscellaneous bits of stories? The answer: nothing. Yet, Floridi tries to invoke Kant to reinforce a sense that the narrative structuration of the self is feasible after all (i.e., selves, if they are narratives, are coherent and unitary narratives, at least when dealing with healthy, ordinary selves). Kant would argue for a transcendental subject, a sense of self as a unity of the information about the external world, synthesized as epistemic agent and guarantor of its own information.

Problem with Kant and all those that would follow him was simply that being blind to the very processes of production of information (i.e., the trillions of bits of data processing going on every microsecond in the brain) he assumed he had sufficient information available to him to explain what was going on in the narrative or informational self to make it a unity. But he was wrong. Even Floridi is stymied by this and asks: “granted that the unity of the narrative or informational self and (perhaps) its crucial role in the delivery of a coherent experience of the world must be conceded, what generates it and keeps it together?” The brain itself of course, but we have no direct access to these processes through philosophical reflection or recursive reason to tell us even this much, so we assume we have all the information needed to begin to formulate a solution when in fact we’re caught in a trap laid by the brain itself long before we begin to reflect.

Floridi even admits as much telling us he left such questions unanswered in the past. Instead he defended a notion of self as information arguing that each self should be “conceptualised as being constituted by its information, thus understanding a breach of one’s informational privacy as a form of aggression against one’s personal identity” (Floridi, 11). But if as Scott in his BBT Theory (Blind Brain Theory) tells us we are constituted by a minimal set of data which is both temporary and non-permanent:

If the lack of information, the ‘superunknown,’ is what limits our ability to cognize nature, then it makes sense to assume that it also limits our ability to cognize ourselves. If the lack of information is what prevents us from seeing our way past traditional conceits regarding the world, it makes sense to think it also prevents us from seeing our way past cherished traditional conceits regarding ourselves. If information privation plays any role in ignorance or misconception at all, we should assume that the grandiose edifice of traditional human self-understanding is about to founder in the ongoing informatic Flood…

So instead of having the information needed to conceptualize a self as unity, we are actually informationally deprived and lack all but the minimal set of data that the true chariot underwriting the informational cosmos of the self is the brain of which we have little or know informational knowledge through our recursive system of reason and reflection. Yet, Floridi, admits as much saying clearly, “it is going to be hard to tackle a problem that Plato, Hume and Kant left unsolved” (Floridi, 11).

In the final phase of his essay Floridi proposes what he terms a figural three-phase approach, one that will help him support his pet theory concerning ICTs (Information-Communications Technologies). First, he sets out that basic information theory that will support what he terms the “reconciliation hypothesis” (I quote at length):

In the same way that organisms are initially formed and kept together by auto-structuring (i.e., auto-assembling and, within the assembled entity , auto-organising) physical (henceforth corporeal) membranes, which encapsulate and hence detach (bear with me, more on this below) parts of the environment into biochemical structures that are then able to evolve into more complex organisms, selves too are the result of further encapsulations, although of informational rather than biochemical structures.(Floridi, 12)

The self emerges as a final phase in which the basic mechanism of encapsulation, detachment and internal auto-organization or the pre-ceding. In his words, the self emerges  “as the last step in a process of detachment from reality that begins with a corporeal membrane encapsulating an organism, proceeds through a cognitive membrane encapsulating an intelligent animal, and concludes with a consciousness membrane encapsulating a mental self or simply a mind” (Floridi, 13).

This three-phase structuration process or what he calls the 3Cs entails the flow of organisms that begin in a pre-biotic and pre-communicative environment and enters the corporeal membrane phase that allows the separation of the interior of a cell from the external world. “This is the ontological function of the membrane, as a hardwired divide between the inside, the individual biotic structure, and the outside, the environment” (Floridi, 14). The next phase entails a movement from pre-cognitive to post-cognitive systems once data become encodable resources exploitable by organisms through some language broadly conceived (Floridi, 15). It’s in this phase that the a cognitive membrane is conceive brining about the encapsulation of data for processing and communication. And, the final phase,  represented by the evolution of the consciousness membrane. “We move from pre-conscious (aware) to post-conscious (self-aware) systems once data become repurposable information, including conventional meanings” (Floridi, 15).

Ultimately each membrane phase by phase reinforcing, bonding force from phase one biochemical bonds, to phase two with mutual information or the interdependence of data, and phase three, with semantic bonds that allow the self a stable and long-lasting detachment from reality (Floridi, 15). Ultimately, in that last phase each individual self enters a semantic community of other selves, a social or collective agent system, and is embedded in a distributed network of informational practices:

The I emerges as a break with nature, not as a super connection with it. Such an “unnatural” break requires a collaborative and cumulative effort by generations through time. No individual can successfully rely just on a private semantics (what Wittgenstein calls private language). This is why a single human being needs to be embedded, at a very early stage of development, within a community, in order to grow as a healthy self-conscious mind: mere corporeal and cognitive bonds, in one-to-one interactions with the external environment, fail to give rise to, and keep together, a full self, for which language, culture and social interactions are indispensable. The problem of the chariot therefore may be solved only by taking into account all the bonding forces – physical, cognitive, and semantic – that progressively generate the unity of the self.  (Floridi, 16)

Floridi now shows how ICTs have effected radical modifications across the board as they have become intermediation devices and technologies of the self. As he states it ICTs have induced radical modifications (a re- ontologisation) of the contexts (constraints and affordances) and praxes of self- poiesis, by enhancing the corporeal membrane, empowering the cognitive membrane, and extending the consciousness membrane (Floridi, 17).

Our culture is so enamored to the science fictional progression of the sciences in the NBIC technologies that over the years we’ve come to be so imbued with informational concepts, we even find the very idea of “eterobodiment of the self, or the self as a cross-platform (not a-platform) structure, perfectly conceivable, witness the debate about mind uploading and body swap in the philosophy of mind” (Floridi, 17). We can see this in many of the notions of transhumanist and posthumanist discourse with both visions offering some sense of transgressing our limited natures and evolving into either more genetically superior species or into cyborgs and robotic informational inforgs.

In fact Floridi envisions a time when inforgs will become new type of self, that is, a living organism self-aware of its own information processes (e.g. you) and its own presence within them, and thereby can choose where to be informationally at any point in the timespace continuum. This brings to mind such things as the avatars or telepresence in which scientists use Na’vi-human hybrids called “avatars”, operated by genetically matched humans to explore the planet of Pandora. The only difference between uploading one’s self to a metalloid robotic creature and the avatar is the almost Matrix like sense of being jacked in and teleported into alien flesh which seems to be without self or personality, a sort of lifeless body awaiting the personality and memories of the actual human. At the end of the movie the impossible happens and the quadriplegic human character, Jake, suddenly breaks free of his own body and permanently enters into the alien flesh of his avatar by the miraculous power of a specific sentient Tree that is the central agent and power of the indigenous people of the planet. We get the feeling that the tree is some kind of organic information processing system, an agent in its own right that has taught and guided these peoples over the centuries. One gets the sense that the Tree is itself a sort of Information-Communications Technology:

The Pandoran biosphere teems with a biodiversity of bioluminescent species ranging from hexapodal animals to other types of exotic fauna and flora. The Pandoran ecology forms a vast neural network spanning the entire lunar surface into which the Na’vi and other creatures can connect. The strength of this collective consciousness is powerfully illustrated when the human invaders are defeated in battle by the Pandoran ecology, after the resolute Na’vi were nearly defeated. Cameron utilized a team of expert advisors in order to make the various examples of fauna and flora as scientifically feasible as possible. (see Fictional universe of Avatar)

Yet, there are other types of narratives like Richard K. Morgan’s three volume series about a multi-lived contract killer, an EX-U.N. envoy, lives in a future where selves are no longer bound to singular bodies, but are stored as memory systems implants at the base of the brain and spine that can be implanted, detached, and stored for future use in iterable units across the galactic empire. In his first novel he gives us a taste of this transition:

Coming back from the dead can be rough.

In the Envoy Corps they teach you to let go before storage. Stick it in neutral and float. It’s the first lesson and the trainers drill it into you from day one. Hard-eyed Virginia Vidaura, dancer’s body poised inside the shapeless corps coveralls as she paced in front of us in the induction room. Don’t worry about anything, she said, and you’ll be ready for it. A decade later, I met her again in a holding pen at the New Kanagawa Justice Facility. She was going down for eighty to a century; excessively armed robbery and organic damage. The last thing she said to me when they walked her out of the cell was don’t worry, kid, they’ll store it. Then she bent her head to light a cigarette, drew the smoke hard into lungs she no longer gave a damn about, and set off down the corridor as if to a tedious briefing. From the narrow angle of vision afforded me by the cell gate, I watched the pride in that walk and I whispered the words to myself like a mantra.

Don’t worry, they’ll store it. It was a superbly double-edged piece of street wisdom. Bleak faith in the efficiency of the penal system, and a clue to the elusive state of mind required to steer you past the rocks of psychosis. Whatever you feel, whatever you’re thinking, whatever you are when they store you, that’s what you’ll be when you come out. With states of high anxiety, that can be a problem. So you let go. Stick it in neutral. Disengage and float.

I came thrashing up out of the tank, one hand plastered across my chest searching for the wounds, the other clutching at a nonexistent weapon. The weight hit me like a hammer, and I collapsed back into the flotation gel. I flailed with my arms, caught one elbow painfully on the side of the tank, and gasped. Gobbets of gel poured into my mouth and down my throat. I snapped my mouth shut and got a hold on the hatch coaming, but the stuff was everywhere. In my eyes, burning my nose and throat, and slippery under my fingers. The weight was forcing my grip on the hatch loose, sitting on my chest like a high-g maneuver, pressing me down into the gel. My body heaved violently in the confines of the tank. Flotation gel? I was drowning.4

This notion that we are just stored information, bits of memory that can be re-implanted, updated, extracted, and stored for future systems recovery agents you might inhabit is part of this brave new world we’re plunging toward. All fantasy? Maybe, who knows. We have prophets of the Singularity everywhere these days, but that’s another tale.

Floridi himself provides us a more sober, cautionary tale, reminding us that maybe we might not want to store all those memories, that maybe those memories would lock us down, provide others with too much information, control, knowledge, etc. As he tells it themore memories we accumulate and externalise, the more narrative constraints we provide for the construction and development of personal identities” (Floridi, 18). He goes on to say:

Increasing our memories means decreasing the degree of freedom we might enjoy in defining ourselves. Forgetting is also a self-poietic art. A potential solution, for generations to come, is to be thriftier with anything that tends to fix the nature of the self, and more skilful in handling new or refined self-poietic skills. Capturing, editing, saving, conserving, managing one’s own memories for personal and public consumption will become increasingly important not just in terms of protection of informational privacy, but also in terms of construction of one’s personal identity. The same holds true for interactions, in a world in which the divide between online and offline is being erased. The onlife experience does not respect dimensional boundaries…

Floridi also describes the notion of the digital gaze. We’re all familiar with the Lacan’s notions of the mirror stage as it grew from his early writings in which it was part of the infant’s recognition scene involving the consciousness of himself as a person in the mirror, as an identity. He’d later on add to this the structural imposition of the process of subjectivation in which the Ego is the product of misunderstanding – Lacan’s term “méconnaissance” implies a false recognition. Additionally, the mirror stage is where the subject becomes alienated from itself, and thus is introduced into the Imaginary order. The Imaginary order is the formation of the ego in the “mirror stage”; by articulating the ego in this way “the category of the imaginary provides the theoretical basis for a long-standing polemic against ego-psychology” on Lacan’s part. Since the ego is formed by identifying with the counterpart or specular image, “identification” is an important aspect of the imaginary. The relationship whereby the ego is constituted by identification is a locus of “alienation”, which is another feature of the imaginary, and is fundamentally narcissistic: thus Lacan wrote of “the different phases of imaginary, narcissistic, specular identification – the three adjectives are equivalent” which make up the ego’s history.5

Beyond this physical mirroring gaze lies what Floridi terms the digital gaze, which transfers our sense of self to the online environments within which we interact as what Deleuze and Guattari termed ‘dividuals’ and Floridi terms ‘inforgs’. As he describes it theself tries to see how others see itself, by relying on information technologies, which greatly facilitate the gazing experience” (Floridi, 19). Somewhat like a digital Lacan, Floridi tells continues:

The self uses the digital imaginary concerning itself to construct a virtual identity through which it seeks to grasp its own personal identity (the question “who am I for you?” becomes “who am I online?”), in a potentially feedback loop of adjustments and modifications leading to an equilibrium between the off-line and the online selves. The observing is normally hidden and certainly not advertised. And yet, by its very nature, the digital gaze must be understood both as an instance of presumed “common knowledge” of the observation (“I know that you know that I know etc. … that this is the way I am seen by you”) and as a private experience (it is still  my seeing of myself, even if I try to make sure that such seeing is as much like your seeing as I can). (Floridi, 19)

This new form of self exploration through an intermediary system: third object (Harman),  hyperobject (Morton), machinic being (Bryant) in the process of self-amplification and augmentation is altered and takes on a life of its own separate from one’s on physical gaze. The digital gaze as Flordi reminds us “may become mesmerizing: one may be lost in one’s own perception of oneself as attributed by others. And finally, the experience of the digital gaze may start from a healthy and wilful exposure/exploration by the self of itself through a medium, but social pressure may force it on selves that are then negatively affected by it, leading them to re- ontologise themselves eteronomously” (Floridi, 20).

Finally, Floridi, would remind us that the Self is a negentropic machine: “Selves are the ultimate negentropic technologies, through which information temporarily overcomes its own entropy, becomes conscious, and able to recount the story of its own emergence in terms of a progressive detachment from external reality.” (Floridi, 21)

One of the problems I see is that it cannot account for its temporary emergence as an agent of the brain either, it lacks the necessary information to do that and the information it does have is minimal and based on informational decisions within the brain’s inaccessible regions that we do not and may never have access too. So we sit questioning this minimal informational system as if it were some permanent structure when in fact it is but the flotsam and jetsam of specific activities in an ongoing process of the brain’s on activities that we are essentially unaware of. Ignorance is bliss – don’t they say? Maybe in this instance we hope those neuroscientists actually do discover how the information pipeline connects up, then we might just discover what the self lacks in its estimation of its own identity. But that, too, is another tale.


Adding R. Scott Bakker’s response to Floridi’s conceptions of Self and the 3C’s (i.e., see comment section of The Metacritique of Reason):

Floridi is still trapped in a couple of big ways, I think. The most obvious ways lies in his semantic account of information, the need for information to be genuinely ‘about’ something to count as ‘informative.’ Thus the strange need for something to always be selfing in individuals: it’s not enough to simply be a brain possessing different socio-cognitive systems cued by various information environments, the brain always needs information somehow intrinsically about itself as a ‘self’ (corresponding to our metacognitive intuition of self) which can serve as the unifying, organizational substrate of behaviour – to be the ‘doer’ of those things we think selfs do. On BBT, of course, the ‘self’ of philosophical reflection is simply a theoretical interpretation of a cluster of metacognitive illusions, something that only ‘exists’ when you look at it – kind of post hoc judgment we make regarding ourselves from time to time on the basis metacognitive hints and neglect. This is why they tend to be so horribly convenient (why we can be the self we need to be in so many different circumstances) – namely because it consists in nothing more than packaging the same dispositional swamp for different kinds of downstream social consumption.

Floridi’s ’3 C’s’ three levels of functional encapsulation, each effecting some kind of discontinuity between a system and its environments, is a clear cut attempt to render metacognitive incapacity (the way medial neglect forces the brain to cognize itself in discontinuous terms) into a kind of biological achievement. The self, instead of being a kind of toolbox to be rooted through as occasion (problem-ecology) demands, becomes a ‘virtual machine’ that somehow (because he completely elides the problem of meaning (in this paper at least)) constitutes all the (supernatural) hanging efficacies we assign to the first person.

On BBT, metacognitive encapsulation simply marks the boundary of information availability, and thus a way to track the role neglect plays in metacognition more generally. Our intuitions of personal identity (which so famously confound naturalistic explanation) are artifact of ‘asymptosis,’ the way the longitudinal dimension of the brain’s functioning utterly escapes metacognition and so does not exist for metacognition, leading to the paradoxical intuition of some timeless ‘I.’

1. The Informational Nature of Personal Identity, Minds & Machines., 21.4, 2011, 549-566. (“The original publication is available at”)
2. Wooldridge, Michael (2011-08-24). An Introduction to MultiAgent Systems (Kindle Locations 201-205). Wiley. Kindle Edition.
3. Hume, David (2008-12-11). Works of David Hume. A Treatise of Human Nature, An Enquiry Concerning Human Understanding, An Enquiry Concerning the Principles of Morals, The Natural … Dialogues Concerning Natural Religion (mobi) (Kindle Locations 15482-15486). MobileReference. Kindle Edition.
4. Morgan, Richard K. (2003-01-01). Altered Carbon (Takeshi Kovacs Novels) (Kindle Locations 150-155). Random House, Inc.. Kindle Edition.
5. Jacques-Alain Miller ed., The Seminar of Jacques Lacan: Book I (Cambridge 1988)



2 thoughts on “Multiagent Systems: Plato’s Chariot, AI, and Self in Information Philosophy

Leave a Reply

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

You are commenting using your 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