Monday, October 19, 2020

Affective Politics

Source: Casper Hesp1

It is generally assumed that, developmentally, life proceeds from the bottom up. One of the first tasks for any organism is learning how to anticipate and synchronize it's biological processes with the environmental milieu in which it exists so that it can survive and achieve allostasis. The physiology of an organism is essentially the representational/generative model evolved for this purpose. These models produce affective/aesthetic experiences when sampling the environment for valuable "model evidence". (Through conceptual construction, we call experiences with high arousal/valence the feelings of love, beauty, etc.). The logic of teleology, because it is goal oriented, reverses this order and proceeds from the top down. It begins with a singular truth: our awareness of beauty and love, which we experience under specific conditions. This affirms or "self-evidences" the existence of creatures, like us, who are capable of realizing these states. How are these affective states/conceptual constructions realized? By being embedded in a caring community through which we receive care, extend it to others, and create the conditions for greater enjoyment and satisfaction in living. How is that done? By sharing appropriate models for learning how to synchronize our needs with each other and the environment around us. This explanation integrates our experience with our embodiment and feeling with cognition. 

If this is a fair portrayal of life and love, so far as that goes, then we may be able to derive several important implications from it. The importance of synchrony is widely recognized. Karl Friston and his many research collaborators have perhaps done the most detailed work related to showing how synchronization, and therefore allostasis, can be achieved. They note the highly recursive, even fractal nature of generative models (i.e. relational systems) used by agents, across many scales, for this purpose. Essentially this is a system of models embedded within models, in a seemingly infinite recursion whose limits are unknown. And at any given scale, these models [and scripts] interact with their conspecifics at the same level, providing a wide range of counterfactual scenarios and comparative modeling and policy analysis possibilities. The image of the Julia set2, a famous fractal, has been used metaphorically to illustrate this horizontal and vertical depth. Lisa Feldman Barrett finds that the notion of "constructionism" is very useful when describing how anything, whether emotions, models, or other social constructs are discussed. In fact, yesterday she published a blog post in which she noted "The physical consequences of some of the world's most challenging social problems - poverty, racism, and religious strife, to name a few - could in principle be eradicated by a massive, collective change of mind." Why such a bold statement? Because the way in which we construct our social reality shapes these problems far more than the bare physical conditions and limitations of our environments. To wit, these are problems we both created and have the power to eliminate. 

I hope it is now clear that if we understand the models we use, both those we construct and those we are embedded within, and if we understand how to work with those which are most capable of preventing such avoidable problems, then more people will be able to have satisfying lives filled with experiences of enjoyment, love, and beauty. This might sound simple so far, but experience has shown that it is no easy task. We know that globally, individualist social models are on the rise (even in countries like China where role-based ethics have strongly influenced the culture). The contemporary zeitgeist is antagonistic toward self-control and restraint of any kind. Yet, the constraints at work within models are among their greatest assets, and enable them to synchronize with other models in an ecosystem of relationships. In nature, circadian rhythms constrain the periods during which organisms are active, and dictate when they should rest. Lunar cycles and the seasonal changes in weather over the course of a year constrain entire global ecosystems, and entrain them to a single global respiratory beat, visible in the Keeling curve. The human ability to circumvent some of these constraints must be used with great caution. Artificial light can lead to sleep deprivation, over abundant food sources (especially when refined and over processed) can lead to obesity and a wide range of other metabolic diseases. We most emphatically need to understand and use models to constrain ourselves, and recognize that those which remove too many limitations may lead dangerously close to asynchronous behavior that is incompatible with individual health and, at larger scales, a sustainable planetary metabolism.

Source: Karl Friston2
The process for selecting and implementing models that are suited to the sort of lives we want for ourselves, after decades of unrestrained global growth, may need to reckon with a society in withdrawal. How do you cut off the "high calorie" diet of fossil fuels and the luxurious and wasteful lifestyle it enabled without confronting this? For some, sobering up from this bender will be very painful. As Franklin Leonard said "When you're accustomed to privilege, equality feels like oppression." This insight applies to synchronization with ecological patterns just as well as it does to discrimination based on race or class. It's going to take a careful and steady application of everything we know about ourselves and the world we live in to repair our relationships and return to healthy lifestyles. It is also hard to overcome the fear that emerges from inaccurate modeling processes. Even though we may know they are an unreliable guide for future behavior, we've become habituated through repeated use. What we do know, however, suggests there is not just one, but many alternative paths capable of providing a standard of living above that of the present (just as there are also many paths to a declining standard of living). When you ask a child what they want to be when they grow up, they tend to only respond with models that they have been exposed to already. Imagining a way of living that they have neither practiced, seen, nor been told of can be very difficult. 

Zen Mind

John Daido Loori, in "The Heart of Being: Moral and Ethical Teachings of Zen Buddhism", remarks that from the perspective of the Buddhist patriarch Huineng (638-713), all things are nothing but phenomena in the mind. Reality is created with our minds. In the coming together of the organ of perception, the object that it perceives, and consciousness, reality is established. His successor Shenhui (684-758) wrote two frequently quoted Zen verses that describe the mind as "prior to heaven and earth, formless and unchanging, yet with the power to control all the changing things". Wing-tsit Chan, in "A source book in Chinese philosophy", wrote that the standard sayings within Zen are "point directly to the mind" and "see one's nature and become a Buddha". The effect is to reduce Buddhism to a concern with one's mind alone; everything else is unnecessary. This sort of nondual “mind only” idealist perspective explains a lot about those to whom it appeals. Andy Clark once told Karl Friston that he was like the austere philosopher W. V. O. Quine, who had once said, of what he considered to be an unnecessarily complicated idea, that its “overpopulated universe is in many ways unlovely. It offends the aesthetic sense of us who have a taste for desert landscapes.” However given Friston's frequent use of fractal descriptions, that may not be entirely accurate. Zen, with its rock gardens and minimalist appearance, is certainly no stranger to austerity as well, and likewise, fractal metaphors abound when describing dependent origination. Aside from this shared aesthetic sense, the neuroscientist Friston and the patriarch Huineng are both, first and foremost, focused on how the mind operates. Perhaps the history of Zen is not unlike an attempt to understand the mind as a sort of "generative model”, where afflictions result from inaccurate modeling, dharmas are more accurate apprehensions of cosmic order, and Buddhahood is allostatic balance. It's an analogy that doesn't come without qualification, but may be a useful one nonetheless. In the Platform Sutra of Huineng he describes the four great vows:

Julia set
We vow to deliver an infinite number of sentient beings of our mind.
We vow to get rid of the innumerable defilements in our own mind.
We vow to learn the countless systems of dharma of our essence of mind.
We vow to attain the supreme Buddhahood of our essence of mind.

Huineng goes on to say (paraphrasing): “Who are these sentient beings within our mind? They are the delusive mind, the deceitful mind, the evil mind, and suchlike minds, all these are sentient beings. Each of them has to deliver themselves by means of their own essence of mind, by means of right views. Let the fallacious be delivered by rightness, the deluded by enlightenment, the ignorant by wisdom, and the malevolent by benevolence. As to the vow "We vow to get rid of the innumerable defilements in our own mind," it refers to the substitution of our unreliable and illusive thinking faculty by the wisdom of our essence of mind. As to the vow "We vow to learn the countless systems of dharma," there will be no true learning until we have seen face-to-face our essence of mind. As to the vow "We vow to attain supreme Buddhahood," when we are able to bend our mind to follow the true dharma on all occasions, then we may consider ourselves as having attained Buddhahood.” We can review this in light of the preceding section. Healthy generative models are essentially purpose built for realizing synchrony, and we each experience the evidence of their success or failure in our affect. Our affective experience, in turn, informs (and is informed by) our relationships with each other, including political and economic decision making. Indeed, we must be able to recursively modify our models to be in synchrony with each other and the environment. So in addressing the hearts and minds of the people, here Huineng is demonstrating that he understands this to be the appropriate level where change must begin.

Notes:
[1] This figure appears in "Deeply Felt Affect: The Emergence of Valence in Deep Active Inference" by Hesp, C., Smith, R., Parr, T., Allen, M., Friston, K., & Ramstead, M. (2019, December 3). Caption reads (paraphrasing slightly): "A schematic breakdown of the nested processes of Bayesian inference in terms of the affective agent presented in this paper. At each level, top-down prior beliefs change along a gradient ascent on bottom-up model evidence, moving the entire hierarchy towards mutually constrained posteriors. Perception (light blue) provides evidence for beliefs over policies (blue) and higher-level contextual states. Action outcomes inform subjective fitness estimates through affective charge (brown), which provides evidence to inform valence beliefs (orange). These nested processes of inference unfold continuously in each individual phenotype throughout development and learning. In turn, the reproductive success of each phenotype provides model evidence that shapes the evolution of a species."
[2] This figure incorporates the Julia set. It appears in the papers "A Free Energy Principle for a Particular Physics", "Parcels and Particles", and "Neural and phenotypic representation under the free-energy principle". But also see "Answering Schrödinger's question" for an earlier version of the figure where a Romanesco broccoli was used in place of the Julia set.

2 comments:

  1. Do you really think Cindy Crawford and other models are worthy to follow?

    ReplyDelete
  2. Your double entendre here is clever - Cindy Crawford has had an iconic career as a fashion model. That's a different usage of the term model. Within the field of active inference the value of a generative model is determined by its ability to realize synchrony with its environment. This is entirely context dependent. The stereotype with fashion models is that they "look good on the outside but may be empty on the inside." Generalizations of this type are likely wrong when applied to any individual. But let's extend this analogy. Should we be concerned that we might adopt a social model only to find out later that, though it appeared fit for purpose, it was completely inappropriate? Absolutely. Worse still, could an apparently good looking social model harbor within it something that only later we discover is actively antagonistic to social harmony? Yes. Not only could a model be "empty on the inside", it could be actively destructive in a highly deceptive way, like a Trojan horse. In short, if we were to place all models on a spectrum spanning from highly beneficial to highly harmful, very few would be perfectly neutral.

    ReplyDelete