Brain Sciences and Artificial Intelligence
Hegel's Science of Logic (3)
The Science of Logic conversation series is a window into how contemporary thought may intersect with Hegel’s Science of Logic. In the first discussion, Layman Pascal and I discussed his “Metaphysics of Adjacency” (MoA), which brought us towards thinking spaces of closeness (highly sensitive spaces), and in the second discussion, Daniel Garner and I discussed his ideas of “Counter-Enlightenment,” which brought us towards thinking the affirmation of the limitation of reason in order to combat the tendencies of rational totalisation. Hegel’s Science of Logic helps us think both of these dimension insofar as his logic wants our being to get as close to nothing as possible, so that we may mature our approach to concrete becoming, and his logic wants our being to rationalise reason itself as a process of becoming which cannot stand on its own as a fixed totality (or what Garner will call '“autonomous rationality”).
In the third conversation, Thomas Hamelryck and I take a shot at approaching contemporary theories of the brain sciences and machine learning with the aid of Hegelian logic. Believe it or not, we are not the first individuals to consider putting contemporary theories of the brain into conversation with Hegelian logic. We were in part inspired by an article titled “The Dialectics of Free Energy Minimisation” which aims to frame polarisation in the field of brain science theory (between cognitivism and enactivism) as a problem of the very interpretive historicity of the contemporary unified theory of the brain itself.1 This theory is often referred to as the “Free Energy Minimisation Principle”, which deviates from an external model of the brain as a neuronal process changing in relation to external stimuli (“neuroplasticity”), towards an internal model of the brain’s own active self-modelling processes. The brain is recognised as an anticipatory structure engaged in continual processes of self-maintenance and transformation via the active inference of perception-action loops.
The major connection with Hegelian logic here, is that we escape both a scientism that reduces subjectivity to neuronal processes (“neural correlates of consciousness”),2 and we escape a pre-modern philosophy that represents a general science of Being as the Great Outside.3 Instead, we are within a science that has affirmed the Kantian break, introducing the idea of subjective a priori’s and transcendental correlation.4 This basically means that we can no longer understand either a reductionist being from the outside, nor a holistic Being of the outside, but rather have to understand the way in which “the brain” (or “the subject”) is always-already engaged in active processes of internal self-modelling in order to reduce the difference between its actual state and its goal state (ideal internal model). In other words, we always-already approach the Great Outside (external reality) through our internal boundaries or self-models (what are called Markov Blankets in the technical literature).5 Thus, we can no longer think about the brain as neuroplastic change (basically infinite malleability in relation to external stimuli), but rather have to understand the paradox of the brain as only changing to stay the same. Consider that the brain will change in relation to external stimuli, like say, technological accelerationism, but it will only do this so as to keep its internal model as self-similar as possible. If we work through this paradox, what we are dealing with is the paradox of identity and difference, or self-similarity and otherness, which is so crucial for a Hegelian logic capable of containing becoming.
So what that we have to work through the paradox of changing to stay the same? Well, this is important because it pushes science itself to its own limit, i.e. its own internal emphasis on prediction. Ultimately, what is on the line with thinking through the contemporary unifying theories of the brain and Hegelian logic is the dimension of prediction and surprise as it relates to our identity (self-models). Hegel knew that we could never precisely predict the future (we can only interpret our condition retroactively). Science, in contrast, aims to predict everything with minimal error. Consequently, the contemporary brain sciences see our brains (funny enough), as engaged in predictive processing in order to minimise entropy (which is perceived as a proxy for uncertainty, and measured by surprise, i.e. deviation from anticipated outcomes). This tension is fully actual in the contemporary brain sciences, between the aforementioned paradigms of cognitivism and enactivism, because, although our brains always try to approximate a reduction of entropy, this approximation always and necessary fails. Perhaps the right poetic framing of this situation, is that we can never totally banish disorder/chaos from our hearts, since this is our very heart itself.6 In the technical scientific literature, failure is called “surprisal,” but in the logic, may be something close to what Hegel thinks of as negativity (again, as the difference between anticipated outcome and the real). Moreover, the containment of entropy/uncertainty (measured by surprisal), may be close to what Hegel calls Absolute Knowing in the Phenomenology or the unity of Being-Nothing in the Logic. Both Hegel’s Absolute Knowing and his logical unity of Being-Nothing opens a qualitative transition internal to the historicity of science itself, a movement away from “predictive processing” to an exploration of mystery and surprise (or a positivisation of the negativity of reason).
What changes when one contains entropy/uncertainty (surprisal) as the core of one’s own self, is that one is no longer concerned with creating and finalising the perfect model, but rather recognises that one’s own becoming or process is the “perfect model.”7 Practically speaking, one’s “perfect model” as “time itself” or “process/becoming itself” is one that includes within itself, to the very core, negativity (or “surprisal”). Perhaps this reconciles one of the commonly recognised problems with predictive processing, the problem of why we do not simply reduce our existence down to bare repetition? Why do we not isolate ourselves in stillness and darkness until the end (the last surprise: death)? If we were purely predictive processing machines, we would create such an environment of total isolation because we would then be able to attain a (near) perfect prediction. But such behaviour represents an intense extreme of human behaviour: say ascetics or monks isolating themselves in monasteries or the wilderness, affirming (but not containing) the Nothing. In other words, a pure reduction of the brain to predictive processing cannot represent a general theory of how human brains develop processually. In the real of human behaviour, we do not only minimise surprise, but also actively seek surprise (where one’s model can be pushed to its limit, and potentially, transformed by the other).
The paradox is resolved in Hegelian logic by recognising that what underlies the entire phenomenological journey is the path towards Absolute Knowing, where one discards of the perfect model as an other-object to be actualised in space as a complete identity. Consequently, in Absolute Knowing, surprise is no longer perceived as “negative” (interfering with the difference between our actual state and our goal state), but rather perceived as the very opening where new journeys (new stories) can begin. Of course, there is much nuance here which probably requires an entire book to fully understand, and years of research at the intersections of the brain sciences and philosophy. But the main point is that we do not want to predict everything down to its bare minimum. What can be handled cognitively via predictive processing, will be the role of artificial intelligence (disembodied/autonomous rationality). Human beings, or Spirits unfolding their Concept, will be opened to a new form of cognition, which will be much more open and speculative, mysterious, involving much surprise. What we really want is to unfold our notion and become what we ideally are in the real (where we should be ready to be surprised by our own desires). Consequently, this requires confronting several dimensions we often interpret to be negative, including letting go of old beliefs, losing what we once cherished, and recognising that the true is more complex than our rational models interpreted through a spatial lens, often via low-level cognitive mimicry of the other’s desires.8
These are some of the major scientific questions that we seek to further explore in the Science of Logic course. If you want to learn more, visit Philosophy Portal. The Science of Logic starts January 16th 2023 (pre-course sessions are also being offered), and features guest lectures from Layman Pascal, Daniel Garner (of O.G. Rose), Alex Ebert, Thomas Hamelryck, William Rupush and Thomas Winn. Together, we hope to demonstrate the ways in which Hegel’s Science of Logic speaks to our present in a way that we can rethink everyday life, creativity, technology, and science.
This polar opposition oscillates between cognitivists who have a more secluded/isolated understanding of the brain, and enactivists who have a more open/connected understanding of the brain, see: Boonstra, E.A. & Slagter, H.A. 2019. The Dialectics of Free Energy Minimisation. Frontiers in Systems Neuroscience, 10.3389/fnsys.2019.00042. In my teaching of the post-Hegelian excess of Nietzsche’s Zarathustra, we see a “brain” (qua spirit) that oscillates between both positions, seclusion/isolation and openness/connection, towards a beyond of concrete determinate being leading in the void.
I have written about the limitations of this view of the brain here, see: Last, C. 2021. The Difference Between Neuroscience and Psychoanalysis: Irreducibility of Absence to Brain States. Neuropsychoanalysis, 1-12. DOI: 10.1080/15294145.2021.1926312.
Slavoj Žižek makes this point central to his masterwork, Less Than Nothing: “It all begins with Kant, with his idea of the transcendental constitution of reality. In a way, one can claim that it is only with this idea of Kant’s that philosophy reached its own terrain: prior to Kant, philosophy was ultimately perceived as a general science of Being as such, as a description of the universal structure of entire reality, with no qualitative difference from particular sciences. It was Kant who introduced the difference between ontic reality and its ontological horizon, that a priori network of categories which determines how we understand reality, what appears to us as reality.” (p. 9).
Kirchhoff, M., Parr, T., Palacios, E., Kriston, K. & Kiverstein, J. 2018. The Markov blankets of life: autonomy, active inference and the free energy principle. Journal of the Royal Society Interface. 10.1098/rsif.2017.0792.
Karl Friston himself here follows a Freudian model of the psyche, whereby a reduction of entropy is equal to what Freud called the “pleasure principle.” From this foundation, we could reason that “surprisal” is not only similar to what Hegel refers to as negativity, but also close to what Freud referred to as the death drive that disturbs the pleasure principle from within.
Daniel Garner may be pointing to the same thing in his recent article: “Spatial Incompleteness Enables the Fullness of Time” with the “perfect model” representing the “final point of full presence” and the perspective shift on its actuality/impossibility as representing the “temporal completeness.” Here we should note that Hegel himself refers to the state of Absolute Knowing as one that “discards the time form” (with the time form itself structuring the “journey” of the Phenomenology.
On this level we reach the limits of the Oedipal complex as a model for desire, and perhaps have to think the dimension where desire becomes drive.
Beautiful