Blogs

I'm reading Micheal Gazzaniga's book Whos In Charge. Gazzaniga is a neuroscientist who i place up there with Le Doux in terms of important (legible to mere mortals) contributions to the field. Gazzaniga's research work was mostly to do with epilepsy 'split brain' surgery.

I will say that, this isn't a book to read per se, it is more a case of letting it wash over you. It alternates anecdote, hard neuroscience, and age old imponderable philosophical questions.

The first four chapters explore the idea that the brain is, in spite of how much we love the idea of free will, operating completely automatically based on prior programming. It does this by employing a large number of semi-independent modules that rapidly and efficiently operate in parallel to perform an 'automaticity'.

Along the way he references Le Doux's low road procedural mind. Le Doux said in short that everything we encounter is processed simultaneously by a fast 'low road' procedural pathway, and by the slow 'high road' pathway, the latter of which can potentially modify the results of the low road's approach. Daniel Kahneman's system 1 and 2 thinking is an attempt to describe this phenomenon in lay terms, even at the risk of implying that the brain can choose to process either procedurally or deliberatively. Gazzaniga clarifies that the procedural system is the dominant pathway. System 1 is what the brain is doing, all the time, in parallel, at massive scale.

About 85% of the 87 billion neurons in the brain are devoted to procedural processing. And while this is not considered a hard figure, Pragya Argawal in her book Sway showed "that the brain is capable of processing approximately 11 million bits of information every second, but our conscious mind can handle only 40-50 of those bits". This puts my internet mime figure of 95% unconscious, as way too low. If this new data is correct, then another way to express this is that 99.99959091% of all the information that flows into the body from the senses is processed procedurally, unconsciously and automatically. Its why we can arrive at work, having driven the whole way in a day dream.

The interpreter

One of the most profound insights Gazzaniga learned from studying people who have the connection between their left and right hemispheres surgically severed, is the idea of the interpreter.

The interpreter is geographically located in the left cortical hemisphere, but its not considered part of the high road deliberative pathway. It is more of a bridge between the fast procedural pathway (that we use for tying shoe laces), and the slow deliberative pathway (that we use for doing hard math problems).

The interpreter's role is to provide an explanation for the procedural pathway's behavior. The epilepsy patients showed how the brain first acts, then later makes up a story to explain it. Its the output of the interpreter that gives us the impression of conscious control or free will. It makes us feel like we are in charge. So for those of us that have been trying to be more conscious and more present, sorry to say, but Gazzaniga gives us some bad news, that we actually operate first and foremost from force of habit.

Importantly, he makes it clear how finite and limited the recruitment of the cognitive deliberative processing pathway is. Instead of operating in parallel, it functions sequentially, one chunk of data at a time. So using it is orders of magnitude more metabolically expensive. Blood sugar rapidly depletes when it is brought on line. Therefore we can only use it for short periods. If we are aspiring to live each moment in the present, sorry, but again, Gazzaniga suggests that the biological architecture of the brain is not really capable support this except for brief periods. He illustrates this with extensive lab evidence around the rapid deterioration of math performance, and also in real life studies for example in parole boards that only give considered decisions immediately after a meal. The rest of the time they just resort to the default of parole denied.

Nomenclature

I have a desire to establish some orienting naming here. If raw sensory data, arrives and is sent to both the amygdala for first pass actioning, and also to the sensory cortexes for more careful inspection. Then if there is no significant discrepancy, then the procedural low road's decision stands. Otherwise the data is sent for further processing deeper into the cognitive machinery. Either way the interpreter spins up an explanation. When a truly novel situation is encountered, that the brain has no previous experience with, the exact same process occurs. The brain uses the next closest experience from its database to inform both the action and the story. This process is efficient, even if the interpreter's story makes no sense and is verifiably false. BTW in LLMs this is called hallucination.

Now this is all bolted onto the basic structure of learning. Donald Hebb in 1949 showed that learning is implemented by neurons firing, and reinforcing the interneuron connections through each subsequent firing. Then went on to show the structure of multiple dendritic arms to each neuron, that serve to map ideas to other ideas in vast interwoven probabilistic chains. We are born with some templates, and these get further shaped throughout life, and especially when young, in a 'neuroplastic' manner.

These synaptic relationships or 'neural pathways' comprise an association engine. For example that thing on the ground that is long and skinny can be a stick or a snake, that snakes can be dangerous, and sticks are usually not. A leads to B. B causes C. These neural pathways comprise the important prior programming data that the procedural brain uses to make it's decisions. This is strikingly similar to the way that AI large language models (LLM) are pre-trained.

Then when each life situation is encountered, the prediction engine kicks in. LLMs refer to this as inference. All the millions of synaptic firing strengths, serving as weights or probabilities, are tallied up to produce a decision, so as to mobilise resources for a somatic response. Significantly this mobilisation occurs before the story, sometimes many seconds before. If there is a discrepancy or 'prediction error' the process feeds back to the association map. The neural pathway strengths or 'weights' get shifted slightly to incrementally compensate for the error. In this way the brain learns.

But before you get all excited about this, it takes lots of repetitions and a number of other important preconditions before this neuroplastic change occurs in a meaningful way. A few nice experiences is not enough to rewrite the profound trauma of early childhood neglect. I will go more into this in the weeks ahead. Also note that there is an important difference between the brain and AI machine learning. The latter can not yet learn as it goes, it relies solely on pretraining.

So, i think thats all the brain science we need to continue. Three layers, the efficient procedural brain, the expensive deliberative brain, and the story making interpreter.

Lessons and implications

Robert Sapolsky in Determined provocatively weighs in by arguing that all present behavior has deterministic, prior causes. In the hours and days before, my hormones mould my thoughts and emotions. In the preceding months and years, experiences and conditions shaped neural pathways. Further back adolescence and childhood experience permanently shaped my brain at its absolute most plastic state. Going even further back, say my mother had a virus during my foetal gestation, is a predetermined part of who i am today. In the same way all of my ancestors back to the beginning of time, shaped my genes and DNA. None of this i have any control over. Our predetermined behavior and narratives are fast, confident, and often just plain wrong. Shocking i know.

However to my mind the free will argument is a red herring. 'Free will' in the sense of regulated conscious agency, for my own purpose is assumed as a given. My interest in this work is around the illusion of choice we have around procedural behavior. If i respond in conversation in less than half a second, i have inevitably responded from a place of procedure. To assume, in this situation, that i am in any way conscious or 'present' is a fiction and self delusion.

The intention to slow the conversation down enough such as in relational practice, or disciplines like Bohm dialogue or Socratic inquiry, is meaningful so long as we understand that responding more consciously involves very metabolically costly deliberative processes.

If i take Gazzaniga's point that deliberative processing is not just metabolically costly, but the blood sugar needed to power it is literally like a battery, then once that battery is flat, it is flat. So i have to choose my battles, and recruit wise mind cognition minimally and surgically.

From that line of thinking, i'm left with decided tension around our aspiration toward being conscious. In relational practice modalities there is an almost religious devotion to this. It seems to assume that such an unautomatic functioning is not only possible, but that i can learn it, and that it's the only one true way for me to live.

Intuitively i sense our promotion of consciousness is an idealism and a matter of identity. Some sort of attempt to control my way around the fear, uncertainty and messiness of this life. Perhaps to avoid having to look at my wounding. As much as this aspiration seems honorable, the brain does not appear to possess the resources to do it for more than brief periods. An analog is, while it is possible to 'control' my breathing, i can only do so for a brief period. So how do we reconcile this tension?

Deliberative processing is possible, but it's a scarce resource, not the default state. Le Doux, Gazzaniga, and Kahneman are clear, 'ego depletion' is real. We cannot run conscious, effortful attention continuously. The aspiration to be permanently 'present' in the way that we as practitioners frame it is neurologically ungrounded. It's not a failure of discipline, it's a hardware constraint.

If i was to be truly mindful to my entire gamut of fear, alarm and caution that my nervous system has to process to survive every single day, i would probably go insane. Being present to all the reality of my embodied trauma, sounds utterly exhausting, and frankly re-traumatising.

A resolution?

So, i am loathe to leave us stranded with all the horrors of this framing, even if a fuller reconciliation needs another day, and another blog.

However i will say that there is one question worth sitting with. Instead of trying to change them by brute force, "What are my procedures actually doing right now, and are they serving me?".

I wonder if the goal isn't continuous deliberative processing, but gaining some ability to notice when my behavior is a product of procedural mind. This is the kind of meta-awareness that both relational practice, and nervous system models like PVT conceive, and which feels more learnable.

This aligns with what Ellen Langer says in her book Mindfulness. That mindfulness in the more conventional buddhist - Kabat-Zinn sense is where i am quieting the mind, and quieting judgement. Whereas Langer's version is kind of the opposite, using the (deliberative) mind to actively inquire into behaviors and experiences, and thinking harder to look beneath habitual and premature categorisation. If i was to paraphrase that it's an examination of the interpreter's narrative.

Here, the example of our 'Connect to self' / grounding practice comes to mind. At the start of practice my teachers always did a 4 min guided experience of noticing body sensation. It's not that i need to connect to myself in order to connect with others, it is more just a chance to regularly practice noticing.

I was taught that this grounding also is not about trying to change my experience (because we now know, that is not possible), but it's a practice of noticing what is there. What is the quality of my breath just now? Just noticing, not needing to change it. In hindsight this seems to function in the manner of an audit or inquiry. Wow, my breath is quite shallow and diaphragmatic just now. I didn't notice that until i inquired.

In this light this inquiry probably doesn't require sustained deliberative processing. Just brief instances of noticing. Short and sharp, done immediately upon hearing the invitation. That feels important because it suggests the inquiry might be less metabolically costly overall by virtue of its brevity. I am not expecting to be able to sustain it.

The Connect to self isn't asking me to be present for 4 minutes. It's asking me to make contact maybe a half dozen micro instances during that 4 minutes. Each contact is brief, and the gaps between are just fine. The structure of the guided inquiry is doing a lot of the metabolic work, i don't have to generate the attention from scratch.

There is something else. A world class concert pianist plays from a place of flow state. From a place of procedure. The pianist cant possibly think their way into something that happens so fast and fluidly. The procedure of playing a piano is in fact shaped though thousands of hours of pre-training the procedural 'priors'.

In this sense, an artful conversation can and probably has to be predominantly undertaken procedurally, using a package of skilful procedural capacities. This tells me that not all procedural programming is bad habit, born of fear, or of other unknown provenance. There is also art, wisdom and intuition. So even if i could do so, i do not want to do away with procedural functioning.

Instead i might want to occasionally check that my procedures serve my current values and aspirations. After all, some of these procedures are rather bizarre. Every attractive person of the other gender sets off a dopamine cascade designed to propagate the species. The same for every cake and cookie's urging to put on fat for the winter. With central heating that might be a winter that simply never comes.

Ok so these visits and mini inquires to notice the nature of procedural processes, might allow us to decide if we want to work on reshaping the procedures or not. How we do that is a very big subject, which i leave til another day, but in a single word, it' looks very much like practice is the key.

Say we notice that we have a habit of interrupting. Therefore we design a structure that reliably generates new experiences where not interrupting feels good. This avoids us trying to stop interrupting in the moment through will power alone.

The lessons i am taking away are these:

• the goal isn't to be less procedural, its to have better procedures
• training good procedures frees up deliberative processing for when its most useful
• use deliberative processing surgically, especially to audit old procedures
• the self isn't the author of behaviour, it's the author of the story about behaviour
• personal development is story revision, i am not changing who i am, i am updating the interpreters narrative by changing the procedures
• narrative change is real, but it's a side effect of procedural change, not the cause of it.
• in machine learning pretraining and inference are two cleanly separated processes. The brain runs the same processes but as single combined system. Experience and inference updates pretraining priors continuously for next time.
• procedural programming comes from formative development plus years of ingraining, now calcified.
• many repeated or fewer powerful experiences can shift procedural priors
• procedural prediction error when large enough or repeated often enough can trigger a significant procedural weight update.

Do we realise how profound this model is to our practice imperative?