Anil Seth’s Faraday Lecture has generated a fair bit of chat. I was initially a bit too unimpressed, not enough originality – probably turned off previously by his use of “hallucination”. In fact he is synthesising many of the same sources I am. Nothing new under the sun again (from my perspective).
The notes below are raw from watching the lecture in real time.
Since then, I’ve also skim read his “Being You” (a 2021 book) and it confirms that same synthesis and more – IIT and Tononi; Panksepp and Solms as well as Friston; Dennett right up to date. Different aspects, dimensions, levels and scales of consciousness. Uses as his epigraph, the same Edelman quote from Emily Dickinson as I did many years ago. He was a student of Edelman? … a fellow traveller I’ve overlooked.
A few minor points of disagreement – like treating Nagel’s “like to be” as an necessary component, before he’s said anything about levels – Metzinger but no McGilchrist – but I think we end up with the same model.
=====
Mystery – zzz.
Lost under anaesthesia (main aspect anyway, self-aware aspect – the bits keeping us alive are still there – as as he concludes later – it has much more to do with life than intelligence).
We know it better than anything – a la Descartes
Nagel’s – like to be
Mystery – Chalmers – Hard Problem – zzz
The “real” problem – many different aspects – check
So-called hard problem will “dissolve” – check
Level (how) / Content (of what) / Self (sense of)
Content example “colour” – Visible EM Spectrum
Pink (Lilac chaser) dot illusion.
Prediction machine – check (+ Plato’s Cave)
More expectation / illusions
Bayesian predictions / expectations – check
von Helmholtz process / result (not explicit calculation, certainly not bottom-up from sum of all inputs – nope – high level prediction compared with any available inputs at any level. The game is minimising the unexpected. “Predictive Processing” – generative
Wm James on the case
Pareidolia – hallucinations – hmmmm.
Hallucination as uncontrolled perception
Perception as controlled hallucination.
Kinda – check (OK see intro)
Perceptual echo-chambers (That dress)
The Perception Census
Humility
Self <> World
Self is our perception of “us perceiving”
Bodily, Perspectival, Volitional, Narrative and Social – Self
Bodily ( more Sacks / Damasio stuff) – the “Rubber Hand”
Interoception
Emotion (after James again)
Control (Cybernetics)
ACTIVE INFERENCE – Friston!!!
ROSS ASHBY – system regulation
FRISTON FREE ENERGY of living systems
Interoceptive Predictions
Good or Bad affective experience
Descartes Error!
Part of Nature not Apart from Nature
AI & Consciousness zzzz
More smart (intellectually) is not more consciousness.
Anthropocentrism / Anthropomorphism.
What a mind is, is not separate from what it does
Less clear line between hardware and software – all wetware
Info independent of embodiment.
Another reason to be wary of computing machine language.
Property of life.
AI will only become Real I when it has Real Life – check.
AI that seems to be consciousness may be a really bad thing for us.
Dennett quote – AI Tools not Colleagues.
Mystery that matters. (Good round-up)
OOOH Friston in the front row.
And Adam Rutherford – Genetic evolution not mentioned (a given, surely, yes says Seth.)
Maturana and Varela. More “systems thinking” autopoesis.
=====
For me:
Frankly, it’s not a mystery, it’s clearly common knowledge.
Person who developed it most from Friston is Solms.
=====