Funnily enough it was hearing Mark Solms talk some years ago that led me to acquire his “Hidden Spring” on the strength of it, with no prior knowledge, even though it was some time before I actually had the bandwidth to read him. Like Iain McGilchrist, his work is now an embedded part of my own research. I’m a fan.
After an exchange on X/Twitter yesterday and today drawing attention to Mark’s work to others debating the limitations of public-knowledge or objective-science when it comes to explanations of consciousness – both its causal power and our subjective experience of it – someone shared a recent BrainLand PodCast conversation with him.
It’s advertised as being “On the neuroscience of sleep and dreaming” (the focus of Mark’s earlier work) but is much wider ranging. As with understanding consciousness itself, so much knowledge of “normal” behaviour and experience is derived from understanding abnormalities and anomalies – the so-called “Lesion Literature” etc. So too with sleep and dreaming. Their differences in relation to waking and non-dreaming experiences, and anomalies in these, are key to understanding their normal reality.
What I and Ken Barrett (the Podcast presenter) both remarked on is the infectious enthusiasm with which Mark naturally covers the range of topics. And he does it from his own immediate experience in research and in relation to the work of others before and since and the reactions of science-politics around these contentious fields then and now. His knowledge is clearly authentic. The honesty of failed hypotheses and unexpected results as part of the process, no need to take credit for the “accidents” along the way. And, as in any other field, it’s much harder to talk if you have to remember hidden or dubious agendas in your story. If you’re dealing with truth, talking under questioning, changing topics as necessary, is free and easy.
Mark Solms knows his and our, brains and minds.
More people should be paying attention to him and his work.
What do you think?
=====
Post Note: Coincidentally also today, I discover Karl Friston – a source in and a collaborator with Solms work – has been added to the participant list at the 2026 HTLGI Conference / Festival at Hay-on-Wye 22-25 May 2026.
Post Note: and just a few days ago, another one I missed – Michael Levin leading a discussion with Solms & Friston and Chris Fields & Thomas Pollak. Not had a chance to view yet. H/T @MacrinePhD
(Very early on Mike responds – in passing – to Mark’s point on Markov Blankets’ purpose dynamically limiting internal and external knowledge, rather than there being an optimum proportion – suggesting maybe an “attractor” defining the proportion or pattern of knowledge between the two? Second mention of attractor ideas in a week. Chris – multiple time horizons in the sub-conscious meta-systems – sounds a lot like John C Doyle – also Tim Kueper’s comment in my earlier Solms piece – how much do we need to meta-know about our knowing whilst we are knowing, and how much of either needs to be conscious.
Mark pretty much stating my own summary of his position:
Consciousness “is” affect.
It’s feeling all the way down.
It’s
‘How do I feel
about what I know
and what, if anything,
should I do about it?’
Lots of good stuff.)
[Post Note: Having mentioned Solms’ “Friston inspired” Free-Energy-Principle and Active-Inference work, another round of “debunking” FEP erupted yesterday on LinkedIn.
Me: When intelligent people start calling each other insane, cultish, zealots, etc or caricaturing each other’s position, I see a lot of talking past each other. The efforts to defend and debunk become equally tedious and noisy.
Me: If people simply want to say FEP, or a particular Pearl or Friston variety of it, isn’t science – of the orthodox, Popperian, falsifiable kind – they’d be right. (And I’ve pointed out myself that there’s been a great deal of hype and overreach associated with FEP and Active Inference. I personally have only a specific narrow interest in it.)
Me: BUT as far back as the “Emperor’s New Markov Blanket” paper, FEP proponents have pointed out that what is being criticised are in fact explicit features. That it has philosophical, even metaphysical departures from orthodox science, by design. Those that can’t see “more than science” will probably have trouble engaging in the dialogue between science and philosophy?
cc: Anatoly Levenchuk
OP: let’s call enron, enron.
Me: [Funny]
Me: Ironic, because Enron has been one of my example cases 🙂Anatoly: Rough decomposition:
- Mathematics / logic / probability: you can think of this as a formal substrate: measure theory, stochastic processes, variational Bayes, information geometry, etc. Not falsifiable as such; It’s a particular representation scheme for certain non-equilibrium steady-state systems with Markov blankets: “you can rewrite their dynamics as minimization of variational free energy / ELBO for some generative model.” FEP is closer to a formalism / representation, not an extra chunk of physics glued on top.
- Process theories built under FEP (predictive coding, active inference, etc.): kind of generative model, form of recognition dynamics, precision update rules, etc. They absolutely can be falsified, compared, outperformed.
- Concrete models and parameterizations: specific task, specific architecture, specific parameter priors → standard cognitive / systems modelling. This is where you plug in data, fit, cross-validate, and decide whether this instance of “FEP-style” modelling is any good.
Use my First principles framework https://github.com/ailev/FPF (load as a file in your favorite LLM and ask about episteme’s representation and “principle to work” chain).
Me: Brilliant summary, thanks Anatoly. Listen-up folks:
“Not falsifiable as such;
It’s a particular representation scheme for certain non-equilibrium steady-state systems with Markov blankets: FEP is closer to a formalism / representation, not an extra chunk of physics.”
That previous “Emperor’s New Markov Blanket” critique.
That “non-equilibrium steady-state” concept is an important one to keep an eye on, later. ]
=====