Content on this page requires a newer version of Adobe Flash Player.

Get Adobe Flash player

Media
Twitter
RssFeed
References

All doing is knowing and all knowing is doing.[i]

Immanence.

Ethology, as defined by Gilles Deleuze in his interpretation of Spinoza’s Ethics, is the study of “bodies, animals and humans by the affects they are capable of”.[ii] Following Deleuze’s idea, it seems that Rodney Brooks is a Spinozist when he draws an ethology of Genghis, one of his robotic Creatures[iii]. “If it starts in the middle of an empty room”, he relates, “he simply sits there. If someone walks up to it, the robot moves away. If it moves in the direction of other obstacles it halts. Overall, it manages to exist in a dynamic environment without hitting or being hit by objects”.[iv]

Spinozian ethology defines a body by its latitude: the catalogue of intensive qualities, dynamic capacities, “…what affects or is affected by the thing, what moves or is moved by it”.[v] He is a Spinozist again when he considers “the problem of building Creatures as an engineering problem”.[vi]

Ethology also defines bodies by their longitude: the total sum of extensive qualities or molecular kinetics, relations of motion and rest, speed and slowness. “Artifice is fully a part of Nature”, Deleuze reminds us, “since each thing, on the immanent plane of Nature, is defined by the arrangements of motions and affects into which it enters, whether these arrangements are artificial or natural”.[vii]

Finally, Brooks is a Spinozist in a third aspect. Whereas Artificial Intelligence has been centred on the idea of representation, of endowing the robot with a global and accurate picture of the world; Brooks’ perspective on “embodied intelligence” is centred on the idea of action in the world. Representation is a plane of transcendence, of organization and development that resides outside, above.

As Deleuze notes, “it always has an additional dimension; it always implies a dimension supplementary to the dimension of the given”.[viii] Spinozian ethology, on the contrary, is about laying a plane of immanence with no supplementary dimensions, “the process of composition must be apprehended for itself, through that which it gives, in that which it gives”.[ix]

Brooks is relaying on this plane when he develops the notion of “an agent being situated in the world”; and Cog, the humanoid in which he is currently working, needs high performance computing for this immanent task! Brooks explains, “it receives a continuous large and rich stream of input data of which it must make sense, relating it to past experiences and future possibilities in the world. It is a participant in this world and must act with appropriate speed and grace.”[x]

Let’s clarify. In the following pages, Brooks’ engineered intelligent Creatures will be used to understand Spinozian ethics as an ethology. Deleuze defines this practice as “a composition of fast and slow speeds, of capacities for affecting and being affected on this plane of immanence or consistency”.[xi] Brooks’ robots are defined by nothing but the extensive, kinetic or engineering relationships of its components (longitude) and its intensive, dynamic or affective capacities (latitude) to perceive and act, to change and be changed by the world which is nothing more than the pure plane of immanence.

In his later work, Brooks has pushed forward these technological developments in order to use them as tools for understanding the nature of human intelligence. He has succeeded by engaging with an array of arguments advanced by the more “ethological” studies in human cognition. In particular Antonio Damasio, Lakoff & Johnson, and Francisco Varela, have developed a novel understanding of human cognition in which bodily experiences, affections and emotions are not peripheral but the core aspect of human reason.[xii]

The following paper will argue that, for being based on embodied affections and perceptive actions in the world, Brooks’ Artificial Life and its related theories of cognition, can teach us a lesson in a Spinozian ethological thinking about human, animal and artificial bodies.

Furthermore, I will conclude by try to link the opposition of representational vs. affective-active cognition with a “higher” Spinozist ethology that characterizes the philosophical work of Gilles Deleuze. On this level, representation in the plane of transcendence becomes Morality; and the affective action in the plane of immanence is a powerful expression of Spinoza’s Ethics.

As we will see in the last part of the text, Deleuze exercised a nomadic form of thinking that is problematic, pragmatic and creative. For both Brooks and Deleuze, knowledge is acquired in a mutually affecting relationship of action with the milieu; this relation is creative because it opens up the door for potentials and events of becoming.

It is not my intention to reduce the differences in kind and context between these two models to a difference in degree or size where Brooks’ embodied intelligence appears as a miniaturization of Deleuze’s pragmatics. My intention is to emphasize a coincidence in cognitive studies and philosophy that is expressed in a movement away from epistemological concerns about truth and representation into a revalorization of knowledge as an effective intervention in the event, in the plane of immanence. Brian Massumi sums up this position: “the question is not: is it true? But: does it work?”[xiii]

Affective Creatures.

Let us consider two thought experiments, from each of them we will try to follow paths that lead to very different approaches to Artificial Intelligence. The first one is called the “imitation game”, it was conceived by Alan Turing in 1950[xiv]. Through two computer terminals, situated in an empty room, you must interact and communicate with two agents situated in a different room.

One of them is a human, the other is a computer; one of them will help you, with its responses, to guess who is who, the other one will try to make you fail. “Your job”, as Katherine Hayles cleverly explains, “is to pose questions that can distinguish verbal performance from embodied reality”.[xv]

The point Turing was trying to make is that, if such a game was made possible and you were defeated in the interrogative investigation, it will prove that machines are able to think. The second experiment is called “brain in a vat”. Imagine a bodiless brain, suspended in a container with nutrient chemicals that keep it alive. A huge amount of threading nerves sway outside of the bucket and the brain is stimulated through them in the same way that it would be if those nerves were connected to the environment through a body.

Would that brain have normal mental experiences? Antonio Damasio offers a very satisfactory response. The brain would not have a normal mind because “the absence of stimuli going out into the body (…) would result in suspending the triggering and modulation of body-states”. The feedback of those body states to the brain, he argues, is “the bedrock of the sense of being alive”.[xvi] If a superbly realistic stimulation was attained, he concedes, the brain could have mental states but that, in fact, would only prove the necessity for “body type inputs”.

Closer to the perspective exemplified in the first experiment we have roboticist Hans Moravec, a representative of the Artificial Intelligence (AI) perspective. This approach describes consciousness as a central system, as an information processing and symbolic software that could in principle be downloaded into a computer.[xvii]

Moravec’s robots, for example, use a central representation system that gathers data from the world in order to situate its position in the abstracted picture of the environment. The second experiment is closer to the Artificial Life (AL) approach which is based on a bottom-up and embodied perspective on intelligence. The research developed by Rodney Brooks and his team, for example, is grounded on the recognition of intelligence as an emergent phenomenon that cannot be understood if broken down into pieces. Their method is to build incremental intelligence based on an embodied interaction with the world.

“When intelligence is approached in an incremental manner”, Brooks explains, “with strict reliance on interfacing to the real world through perception and action, reliance on representation disappears”.[xviii]

This is the reason why Brooks’ developments are mostly based on simpler level intelligences that allow building up the capacities of the Creature and letting it loose in the world with real sensing and real interaction. This is how he reached one of his most famous conclusions: “when we examine very simple level intelligence we find that explicit representations and models of the world simply get in the way. It turns out to be better to use the world as its own model”.[xix]

According to Brooks, the evolution of intelligence took most of the time developing survival capacities like mobility and vision in dynamic environments. This is the essence of being and reacting like a living being, he argues, “sensing the surroundings to a degree sufficient to achieve the necessary maintenance of life and reproduction”.[xx] This level of embodied and affective intelligence provides the ground for the evolution of more complicated forms of intelligence like language and reason.

“To live is to know”. With this aphorism, Humberto Maturana and Francisco Varela tried to convey the same idea[xxi]. In their theory of autopoiesis, the main characteristic of living beings is their capacity to produce the components of their own structure while maintaining their patterns of organization against changes triggered by the environment. In their “biology of cognition” there is no need for a central system and, most important, no need for a mental representation of the world.[xxii]

Early AI was mostly concerned with research on formal systems like games, symbolic algebra, theorem solving, geometrical problems mostly focusing on the development of effective representational systems where the complex features of the world were reduced and simplified by processes of abstraction.

Instead of capitalizing on the singularities offered by an enactive perception of the world, AI systems were mostly driven by representational abstractions that, ironically, were done mostly by the researcher. From the point of view of a Spinozist ethology this is problematic.

As we said before, ethology is the study of the kinetic relations and dynamic capacities that characterize each thing. “For each thing”, Deleuze notes, “these relations and capacities have an amplitude, thresholds (maximum and minimum) and variations or transformations that are peculiar to them”. All these particularities come together in a process of irreducible individuation for each thing or body.

Deleuze suggests that all the singularities that characterize each body (their very own affective capacity) “select, in the world or in Nature, that which corresponds to the thing; that is, they select what affects or is affected by the thing, what moves or is moved by it”.[xxiii]

And he gives the example of the “animal worlds” described by a “Spinozist” German biologist, Jakob von Uexküll. The tick, for instance, has a world of three affects: it is affected by light, by smell and by temperature. This allows him to climb to trees, fall into animals and find blood in the warmest spot of the body. “A world with three affects, in the midst of all that goes on in the immense forest”[xxiv].

These animal worlds, determined by the particular affective capacities of each animal, is what Uexküll called Merkwelt. Brooks is a Spinozist ethologist as well and he also cites Uexküll to object the centrality of abstraction in AI. Abstraction reduces the features of the world in accordance to human affective capacities, so “the program experiences the same perceptual world (Merkwelt) as humans”.[xxv] And this is not what Brooks wants; his ideal is much more ethological for he imagines that “each robot species with their own distinctly non-human sensor suites, will have their own different Merkwelt”.[xxvi]

Once again, Maturana and Varela are very close when they speak about the “cognitive domain” of each living system. Bacteria, for example, have an affective world of “warmth and coldness, of magnetic fields and chemical ingredients”.[xxvii] In their theory of cognition, living organisms are affected only by a fraction of the stimuli that impinges on them. This range of interactions depends on their biological structure.

Because cognition depends on this process of structural coupling, it is not a representation of a world outside but a “bringing forth of a world”.[xxviii] And maybe Deleuze agrees with them when he asserts, “the interior is only a selected exterior, and the exterior, a projected interior”.[xxix]

However, Maturana and Varela understand this as a phenomenology particular to each organism. At this point, it is necessary to establish some important differences between their cybernetic approach and Deleuze’s elaboration of a Spinozian ethology. From the perspective of a Deleuzian ontology fully committed with a realist position that endows matter with creativity independent from cognition, Maturana and Varela’s alliance with a phenomenological tradition is somehow problematic.

In Maturana and Varela’s perspective, it is the organism that “lends significance to its surroundings, creates a meaningful world through its organizational closure, a world that the environment doesn’t possess by itself”.[xxx] What does this means for a Spinozist ethological position? It means denying to the “environment”, in fact, to all the bodies which compose that world, their own kinetic and dynamic capacities, which are creative and meaningful enough in their own right.

The selection performed by the amplitude of affective capacities must not lead us to assume that the world lacks “pre-defined information”[xxxi] as if it was a passive material waiting for the imprinting of transcendental froms.[xxxii]

A second fundamental difference appears between the ethology that we have been describing and Maturana and Varela’s autopoiesis. As we have said, ethology is most of all concerned with affects. As Deleuze have said, “affects are becomings”.[xxxiii] The theory of autopoiesis, however, is firmly grounded on the opposition between autonomy and heteronomy (“invasion” as Pearson describes it), thus, in “a whole conservative metaphysics of living systems”.[xxxiv] An ethology must embrace machinic connections between heterogeneous elements[xxxv]; autopoiesis is mostly afraid of change, it reduces it to a destruction of organization as a sort of transcendent essence that most be protected by the organism itself.

This is where the concept of affection becomes clearly superior to perception or enaction. For Massumi, affect is the angle where “differential emergence” appears.[xxxvi] Affect is a singular or bifurcation point (a turning point) where the “system” opens the door for “mutually exclusive potentials, only one of which is ‘selected’.”[xxxvii] In this sense, it is a transduction, that is, “the transmission of an impulse of virtuality (…) the transmission of a force of potential”. Affect is the moment where a body, by being impinged, is open for an unpredictable non-linear resonation of irreversible change.

Simondon allows us to recognize a dimension that is left outside in the theory of autopoiesis. The relationship between living being and milieu is not one of closure but of openness; the living being not only adapts to the environment by modifying its relationships but also by “modifying itself through the invention of new internal structures”.[xxxviii] By focusing on affection as becomings, as doors to virtuality, ethological thinking appears not as an epistemology but as a micro-ethics of affection within the vortex of change.

Even in this point, Brooks approach stand as an ethology capable of accounting for change as an opportunity. As he explains, “a Creature should be robust with respect to its environment; minor changes in the properties of the world should not lead to total collapse of the Creature's behaviour; rather one should expect only a gradual change in capabilities of the Creature as the environment changes more and more”.[xxxix]

Another ethological dimension of Brooks approach can be found in what he calls “subsumption architecture”[xl]. According to Deleuze, the last dimension of ethology is to study of the compositions into which different things enter to form more ‘extensive’ relations and more ‘intensive’ capacities.[xli] Having studied the affective capacities of things, the idea is to understand how those capacities are enhanced in the formation of more complex arrangements. “How do individuals enter into composition with one another in order to form a higher individual, ad infinitum?”[xlii]

Brooks’ “subsumption architecture”, unlike AI which is based on central and peripheric systems, is based on subsystems or layers. Each layer is independent from the rest and has its own patterns of interactions with the world. “The advantage of this approach”, Brooks explains, “is that it gives an incremental path from very simple systems to complex autonomous intelligent systems”.[xliii] What appears here are “sociabilities and communities”. We attend to a symphony and witness “the composition of a world that is increasingly wide and intense”.[xliv] Affects open the door for becomings and new connections.

Emotional Humanoids.

How complex these arrangements can become? How wider and intense its world can be? This is what Brooks is trying to find out with Cog, an upper-torso humanoid robot currently being developed in the laboratory at the MIT.[xlv] Taking advantage of the computational power offered by recent developments, this project has engineering and scientific goals. The first is to build a “prototype general purpose flexible and dextrous autonomous robot”.

The second is to use Cog as a tool for “understanding human cognition”.[xlvi] In this project, Brooks is taking his approach to its ultimate consequences. The plan is to build a robot that learns to think by engaging it in embodied interactions with a human world. He believes that building a humanoid under the same principles that grounded its previous achievements will finally prove the recent findings in cognitive sciences that are “converging on an anti-objectivist, body-based approach to abstract cognition”.

All these approaches are centred on the idea that intelligence cannot be separated from the process of continuous interlocking between mind, brain, body and environment that constitutes a subjective experience.

Summarizing the cognitive approach behind Cog, we could say that it is engineered against three false assumptions about human intelligence and grounded on four aspects of human intelligence that Brooks and his team consider of central importance. First of all, it is false that humans construct a full monolithic representation of the world. It is also erroneous to think that the brain is a single and unitary control system.

Finally, it is a mistake to think about humans as purely rational and deductive, recent evidence has shown that emotions, metaphors and other “elusive” capacities are central to human rationality.

In order to escape form those false assumptions, Brooks has been engaged with what Lakoff and Johnson call a second generation of cognitive science, characterized by the emphasis put on the idea of an embodied mind.[xlvii] Drawing from current research in cognition, Brooks base his methodology on “four central aspects of human intelligence: development, social interaction, physical interaction and integration”.[xlviii]

Instead of building on the principles of a central control system, Cog is based on an integrative approach where different sensory modalities and motor systems are connected in order to cope with “the multimodal nature of stimuli”[xlix] coming from the world. Development is as important for Cog as it is for a human infant. By building the system as an incremental process, it is possible to achieve learning by engaging Cog with increasingly complex tasks. Social interaction is important for Cog, not only because it allows him to interact with humans but also because it is a grounding mechanism for building more complex behaviours.

Much of the social interaction is handled by a behaviour engine “that integrates perceptions, drives, emotions, behaviours, and facial expressions”.[l] In order to achieve this level of interaction, the laboratory is building Kismet, a head-only robot that serves as a platform to develop behaviours that will be attached to Cog in further steps of the project.[li] The system consists of three drives, three behaviours, five emotions, two expressive states and a diversity of facial expressions.

When Kismet is engaged in social interaction, the input received by a human interlocutor modulates changes in its homeostatic range that in turn trigger expressions and behaviours. This is how Cog will be able to express his internal states. Finally, the embodied dimension of Cog is built upon principles that have guided Brooks’ robotics since the beginning. “Embodiment and physical coupling”, he argues, “allow humans to use the world itself as a tool for organizing and manipulating knowledge”.[lii]

So, for Brooks and his colleagues, the project of building human-like intelligence is inseparable from a human-like interaction with the world. He follows the theories developed by George Lakoff and Mark Johnson, about how the (metaphorical) shapes of thoughts, concepts and language, are based on physical patterns of bodily interactions with the world, body states and neural states.

It is precisely for these last two dimensions, the emotional and the embodied, that Brooks’ current projects still resonate with a Spinozist attention on affect and immanence. Brian Massumi is insistent on the importance of theorizing about the differences between affection and emotion. Following Spinoza, he argues that “affect is an affection (in other words an impingement upon) the body, and at the same time the idea of the affection”.[liii] In a parallel arrangement, mind and body are equally traversed by impulses that become conscious only when an “idea of the idea of the affection” is reached.[liv]

Considering how elaborate Spinoza’s theory of affections is, it shouldn’t be surprising that his ideas are now being taken over by Antonio Damasio, one of the leading figures in contemporary neurobiology.[lv] In his book Descarte’s Error, Damasio elaborates a topology of the emotional dimension of human reasoning.[lvi] The first figure in this topology is constituted by primary emotions, which are basically certain configurations of the body state. They are innate and preogranized, that is, hard-wired in the body. A feeling of the emotion appears as “a realization of the nexus between object and emotional body state”.[lvii]

Feeling is a process of monitoring the connection between changes in the body state (emotions) and the mental images that initiate it. Feelings are qualifiers of stimulus in relation to body states (Massumi also recognizes a process by which affective intensities are qualified but he calls that emotions)[lviii]. Another figure in Damasio’s model is what he calls secondary emotions, where changes in body states are triggered now by mental images. He finally defines a background feeling as the image of our body state between emotions.

A similar diagram of connections is elaborated by Deleuze in his interpretation of Spinoza’s concepts of idea and feeling (affect, affectus), where the former represents a thing or states of things and the latter corresponds to the variation or passage from one state to the other.[lix] Recognizing the discrepancies in the distribution of terms like emotion or feeling, I collected these different versions in order to emphasize their similarities. I believe that all of them coincide in recognizing a process affection that, as Massumi suggest, ascends “by degrees from the concrete to the incorporeal”.[lx]

My conclusion from Damasio, Massumi and Deleuze’s account would be that the idea of “affect” cannot be clear-cut separated from emotions, feelings or even rationality. Affects, as we saw above, are becomings, transducers of potential transformations. As impingements, they resonate through body-states or emotions and qualified feelings of intensity all the way to reason, decision and other capacities that can in turn act back on the body and, through embodied actions, on the environment. Moira Gatens seem to agree when she argues that: “Reason, or the power of thought (…) can not be seen as a transcendent or disembodied quality of the soul or mind but rather reason, desire and knowledge are embodied and express, at least in the first instance, the quality and complexity of the corporeal affects”.[lxi]

To be truly Spinozist means to define the human (and its “higher” capacities) as kinetic and dynamic relations; minds, parallel to bodies, are defined by their affective capacities in a single plane of immanence. They are just two modes of the one Thing, of the one substance or the immanent God… affections of “the fullest and most intense Individual”.[lxii]

Conclusion: Pragmatic Philosophers.

Spinoza’s Ethics is the philosophy of the becoming-active.[lxiii]

Throughout this paper, my intention has been to map a movement in Artificial Intelligence and cognitive sciences within the framework of Deleuze’s elaboration of a Spinozist ethology. The final consequences of a displacement to an ethological perspective can be found in the practice of philosophy. The movement (in all these areas) is from a model of representational knowledge towards a post-representational conception of active thinking. The circle is finally closed, as we found in Deleuze’s own “audacious experiments in thought”[lxiv] the most accomplished realizations of an ethological thinking.

The metaphysical dimension of affect that has been found in our engagement both with contemporary models of embodied cognition and Deleuze or Massumi’s account is that of a powerful resonation across bodies-minds that transmits potentials for transformation, as bifurcation points of becoming, transductions. Affects are of the order of events, we could argue. Deleuze could have agreed with us.

The mind, for him, was “not a space of subjective representation but a sensorimotor interval, a gap that allows difference to intervene between stimulus and response”.[lxv] Affections are the driving force of the cycle around stimulus and response. The intervention of difference is only possible when affections open the door by “its participation in the virtual”.[lxvi]

Deleuze distinguishes philosophy from science and art. The job of science is to produce knowledge, truthful statements about a state of affairs. Philosophy is closer to art in the sense that it must produce sensation or affects more than truths. Deleuze is against a dogmatic image of thought that looks for recognition and develops, as Paul Patton explains, “a non-representational and nomadic conception of thought”.[lxvii]

The only purpose of this practice is the creation of concepts. Concepts are devised for the capture and expression of affections. They don’t have a referent, they don’t provide truths and they don’t reside in a plane of transcendence. What are important to them are their internal “kinetic” relations. Thoughts, like bodies, can be defined by a longitude or extension, a relation of speeds and slownesses between concepts. They have, most of all, a transformative power. Their function, in a first instance, is to create incorporeal transformations, “impose a new set of divisions on things and actions”.[lxviii] But what is the purpose of the creation of concepts?

“Philosophy”, for Deleuze and Felix Guattari, “does not consist in knowing and is not inspired by truth”.[lxix] Philosophy is a form of thinking, not a search of knowledge. Thinking is that “which takes place when the mind is provoked by an encounter with the unknown or the unfamiliar”.[lxx] This encounter constitutes a problem. Deleuze’s idea of philosophy as a form of thinking is, first of all, problematic. Problems, like affections, are forms of events[lxxi], encounters with gates to potential becomings, unthought virtualities.

They don’t open the door for truth but to potential transformations. This is why philosophy as a form of thought is also essentially creative. It shouldn’t be assessed in terms of truth or falsity but in terms of effectiveness. “Thinking”, Patton argues, “is a form of experimentation where the aim is to determine concepts of the events which determine our fate”.[lxxii]

Timothy S. Murphy argues something similar: “Theories are always local and non-representational forms of action”. Concepts, in this sense, are like a “box of tools” and philosophy is an intervention that produces effects and reactions.[lxxiii] It all comes back to the event: affections, problems, concepts, actions. A door is open for the Virtual proper.

“Every concept shapes and reshapes the event in its own way”.[lxxiv] The creation of concepts should be studied by ethology; it is not an expression of knowledge but of behaviours. It is ethical not epistemological.[lxxv]

Ethology, for Deleuze, “is a long affair of experimentation, requiring a lasting prudence, a Spinozian wisdom that implies the construction of a plane of immanence or consistency. Spinoza’s ethics has nothing to do with a morality…” In the plane of immanence, bodies act inside the event and for that event only. Ethics is only concerned with the plane of immanence.

Where Morality offers transcendent generalities (Good or Evil): This IS food, This IS poison; Ethics capitalize on the singularity of the event (good or bad): when, where, how and for whom is this food or poison?[lxxvi] In the plane of immanence, each thing has particular relations and capacities that constitute an individual in an event. No judgment is possible, events escape all generalizations.

A particular encounter of bodies (that themselves are always changing), a super-singular conjunction of bodies in time-space. This encounter is the door for potential transformations, nothing less than becoming.

The only form of intelligence capable of dealing with this dynamic process of becoming is an immanent intervention in the world. “Wisdom” is only attainable by intervening and shaping the event. This applies to all kinds of bodies in the plane of Nature: robots, humans and philosophers.

What is common to all the positions described here, Brooks departure from Artificial Intelligence, Damasio and his work outside an abstract and disembodied cognitive science of a Cartesian tradition, and Deleuze’s attempts to supersede the Kantian tradition, is a rejection of representation in the most broader sense, as a plane of transcendence or organization, including transcendent values and Morality as models of abstraction. “The illusion of values is indistinguishable from the illusion of consciousness.”[lxxvii] All these examples could be understood, within the tradition of Spinoza’s Ethics, as “a typology of immanent modes of existence”.[lxxviii]

 

 

HTML Comment Box is loading comments...