From elementary computation to ethics?
This week, I stumbled over David Chapman’s fascinating reflections on Buddhist morality. He argues that American Buddhist ethics are largely just reflections of modern liberal secularism dressed up on traditionalist garb and augmented with meditation, and then goes on to dismantle the traditional Sutra, while getting caught up in the medieval ethics of traditional Buddhist scripture.
This inspired some local discussions, and prompted me to spell out a quick…
16 step path from elementary computation to normative regulation
The universe is probably a reversible computational process, i.e. a non-branching state machine, where every state is given by a set of bits, and a transition function permutes the bits in a deterministic fashion. (Some alternatives are possible, but seem to be more difficult to get to work while accounting for observable physics.)
This is the lowest layer; here, the universe is causally closed.The systems we observe and interact with are realized at various levels of supervenience. They are not causally closed (i.e. the underlying dynamics of the universe can break their dynamics, so they transition into different systems). These systems are causal systems. In my view, causality is conditional state transition, i.e. a mode of description whereby we identify local properties that predict future local properties.
If 1. is true, then the global, elementary state transitions are unconditional, but of course we cannot build predictive models in that mode, because we do not have access to the global state. The features and objects that we use to make the world predictable are aggregates of available bits that are chosen to maximize predictive power.
A computer is a causal system that can enact algorithms. An algorithm is a set of states and conditional transitions between them, where all the conditions are known. Computers can be probabilistic or deterministic. Deterministic computers can become probabilistic by adding a source of well-distributed unpredictability to their state-transition conditions, and probabilistic ones can approximate determinism by stacking the outcomes.
We are usually interested in irreversible computational processes, i.e. ones that can delete information. Irreversibility allows us to branch and visit states regardless of parts of the previous history. Living systems, control systems, and more generally systems with attractor states are all irreversible. To implement an irreversible system on top of a reversible one, the substrate needs to be open. Openness allows us to build a local system that loses the bits that make it irreversible into the environment. To make openness effective, it does not need to be permanent. It is sufficient that the global system experiences periods of local disequilibrium that make it locally and temporally open. Irreversible dynamic structure requires the existence of such pockets of openness.
The differential between the causal dynamics of irreversible attractor systems and a reversible substrate manifests as entropy. Entropy is the garbage bits that a control system needs to expel to create or maintain order against the disturbances of the substrate dynamics. Expelling information in a reversibly framework also means loss of structure, so the irreversible system needs to replace the expelled disorder by bits that are more ordered than the system itself. This source of negentropy is the energy source that we see as driving the irreversible computation.
Irreversible systems in a reversible universe require a negentropy gradient. This gradient can only be a temporary pocket, hence all irreversible systems are doomed to die, because the universe cannot sustain such gradients indefinitely.
Living systems are adaptable negentropy extractors. The cell is the smallest known universal machine to extract negentropy over a range of variable environmental conditions. A cell is a complex factory to exploit multiple sources of negentropy, to insulate itself from disturbances, to stabilize its dynamics against disturbances that make it through the insulation, and to produce new cells. Its internal dynamics are driven by feedback loops and by the execution of algorithms; these algorithms are mostly encoded in DNA/RNA and accessed and executed in response to meeting suitable environmental conditions. Development and evaluation of the feedback loops and algorithms is mostly performed by mutation and selection.
Organisms are colonies of (mostly related) cells; they are an emergent product of individual cellular dynamics. The main benefit of building such colonies consists in specialization, thereby increasing the range of habitable environmental conditions and the amount of effectively extractable negentropy.
Like cells, organism regulate their dynamics through feedback loops and algorithms, but on an intercellular level. The latter is done in the nervous system, and the most interesting part is the cortex, because it implements the mind, which is (among a few other things) a general function approximator, in the sense of Solomonoff induction. It tries to find the shortest best program to predict present and future environmental states from past environmental states.
Organisms are not the highest level of cellular organization; evolution also runs at the level of families, tribes, groups, eco-systems, organizations and general social systems. Evolution implicitly optimizes the organization of structure for the extraction of negentropy.
The purpose of the mind is the regulation of disturbances that affect the functioning of the organism or of the system it is contributing to. The mind is engaged whenever existing feedback loops and regulation software are insufficient. It will then escalate its modeling to whatever degree is necessary and possible, until it is capable of regulating the disturbance.
The disturbances and the performance of the mind are measured and controlled with a system of rewards and constraints. Because of the mind’s generality, it may find that the easiest way of regulating the disturbances that gave rise to its engagement would be to change the representation of the disturbance. In most cases, this amounts to anesthesia and will not serve the telos of the organism, so evolution has created considerable barriers to prevent write access of the mind to its governing mechanisms. These barriers are responsible for creating the mind’s identification with the rewards and self-model.
When we are concerned about suffering, we are usually referring to disturbances that generate a strong negative reward, but cannot be resolved within the given constraints. On an individual level, disease, mishap and crisis lead to suffering. But we also have a global suffering that is caused by the universal maladaption of humans to present human society, which developed within few generations and deviates considerably from the small-tribe collective environment that we are evolutionary adapted for.
Buddhism, as understood by Chapman, suggests two possible trajectories out of suffering. The tantric way deals with persistent negative rewards by selectively rewiring their sources, and by making the organism more globally functional. The sutric way aims at freeing the mind by removing all sources of rewards. The tantric way keeps the role of the individual within family, society etc. intact, by preserving its mind and rendering it more functional with respect to the organismic telos. The sutric way sees the submission under that telos as morally undesirable, and removes the need of the mind to serve it. As a result, the mind can continue to function as a general computational system, but free of any need, desire or suffering. Its existence or non-existence becomes entirely meaningless. This radical equanimity is reflected in the idea of Nirvana.
The construction of norms starts with a choice between absolute meaning of existence within the given constraints (which leads to systems), nihilism (which, depending on whether you care about innate preferences, either leads to the sutric model or chaos), or waffling (which is what Chapman subscribes to, he calls it “fluidity”, and hopes that it lets him choose the tantra that makes him so happy).
By the way, I suspect that Chapman underestimates the role of the belief in reincarnation for choosing the sutric tradition. If you believe in the unity of all consciousness, you will find that most incarnations suck dramatically harder than the comparatively blissful existence of the prospective student of tantrism (which involves being at the top of the food chain and currently not being in a state of dying or dysfunctional suffering). This creates an imperative to (a) get of of the game, and (b) make sure that your future incarnations do not worsen your chance of getting out of the game (which is much harder when you are reborn as a pig or even a person less prone to Buddhist practice). Incidentally, this also creates a driver for arbitrary virtuous behavior, which makes Buddhism a viable societal mind-virus (i.e. a promotor of useful systemic norms), just like the Abrahamitic religions.
What is moral behavior?
In my view, moral behavior is normative. It is not directed on reward, but on values. Ethical systems are normative architectures to maximize for these values, especially in the face of inter- and intra-individual conflicts between different values, or between values and rewards. Since values are constructs, and there is no known universal objective function for the construction of values, finding the best ethical system is an ill-posed problem. Most working systems are built on an existing partial or total value consensus of a critical mass of individuals, which can then suppress the individuals with conflicting or absent norms.
Various global objective functions have been proposed (utilitarism, desirism, social Darwinism, religious systems) and all have their issues. It seems plausible that in the long run, the prevalence of families of normative systems is going to be subject to evolutionary regulation, which means we should look at sustainability. Sustainability seems to be a good first proxy when designing one’s norms, on the levels of individuals, groups, and globally.
Exclusive regulation by individual rewards (pleasure and pain) is insufficient to achieve sustainability beyond the individual and group level, which is the main error of Ayn Rand’s “objectivism”. (I think she was trying to construct a norm from her own sociopathic intuitions.) The need to be normative (to serve systemic goals regardless of individual reward) is intrinsic to most humans, and results from evolutionary programming. Normative needs are regulated by pleasure and pain, too, but these are directed on transcendental goals.
When playing the organismic game, three strategies are especially salient: Sustainability, in the sense of playing the longest possible game, short-term parasitism (because the current system is doomed anyway, or because we will be ok in the next system too, or because we do ot care), and equanimity (i.e. not playing). There is going to be a tradeoff between the cost and feasibility of external regulation against parasitism, and the benefits of and feasibility to building normative control as parasitism-inhibitory mechanisms into the individual. I currently think that morality should aim at optimizing norms to reflect this tradeoff. But the nihilist attitude, as reflected in the Sutra, seems to be irrefutable.