Machines of Love and Grace - Dr. Karl Friston, Cognitive Neuroscientist

Dr. Karl Friston is the most widely-cited neuroscientist in the world, known for his development of parametric mapping, a method for comparing brain scans between different patients that basically enabled the whole field of functional neuroscience. To follow up his insights into the human brain, Dr. Friston has turned his mathematical mind to other questions of life and consciousness, such as the distinction between living and not-living, the simplest division imaginable between things, and the great challenge of encoding emotional states. We dig into what a soulful robot might look like, and the new class of myths that humans and robots are going to have to write together.

As always, let us know what you think in the comments!

00:00 Go!
00:03:08 What is the simplest division imaginable between things?
00:08:24 Using math for describing emotion
00:16:49 Can you code love into the machines?
00:24:48 Mutual myth-making of robots and humans
00:31:09 The ethics of technological progress
00:39:46 Successes and failures of repairing broken minds
00:47:51 Psychology and human evolution
00:55:16 Is it important to distinguish between the living and nonliving?
01:00:56 Is the origin of life spontaneous generation?
01:09:13 The continuity of beings and planning for an unknown future

#consciousness #artificialintelligence #future


PODCAST INFO: Anastasia completed her PhD studying bioelectricity at Columbia University. When not talking to brilliant people or making movies, she spends her time painting, reading, and guiding backcountry excursions. Michael Shilo also did his PhD at Columbia studying the elastic properties of molecular water. When he's not in the film studio, he's exploring sound in music. They are both freelance professors at various universities.

SOCIAL:

MUSIC:

[optinoptimist]

another great podcast!! great guest, as always, great questions, as always!!

i love the passion yall have for exploring reality, and how you aim to have good, friendly, cooperative conversations, while still really trying to push yourselves, and your guests, past the limits of comfortable truths and convenient narratives, into the realm of uncertainty, reaching for deeper truths to anchor onto in the face of deeper probing into the structure of what is.

i loved the old digital background of space behind yall that you used to have lol, im not one of the modern visual snobs or aesthetic snobs who demand every podcast aim for certain styles or looks if they want to be taken seriously in the "podcast market" lol, that being said, i'm glad you put the curtain up, it's already feeling more like something that's closer to being accepted into the mainstream, which it really needs to be, there's immense value that's being produced, and it's a disservice for us all if it's not on the table of commonly available and consumed content. It's too healthy of a part of our intellectual diet to be left out!!!

i also love the setup with the name under everyone and their work, and the static space background behind everything, and hopefully there will always be some room to maintain the space theme and keep the unique additions.

that being said, i think for the mainstream, you'd probably want to only have the name and work part up from time to time, as a helpful reminder put up here or there when aesthetically pleasing or structurally pleasing, which obviously can feel subjective, but if you survey the "market" (lol), you'll find it's more objective than one might assume, unfortunately lol.

i assume you'll need to fall in line a bit with removing the static space background completely at some point, and keeping the individual guest shots full screen, and the host shots full screen, and the guest/host combo shots fully split screen, and even if you then add a second guest, it seems the preferred aesthetic is often just leaving the grey space grey basically lol, however boring and unintuitive that sounds. for reference i'd posit the TOE podcast when Curt has on multiple guests.

it's upsetting to think that ppl might miss out on this podcast due to the invisible gatekeeping of cultural norms, especially ones as fickle as trending normative aesthetics, but i can imagine a viewer of "popular" science communication podcasts seeing your chill vibe and unique style and comfortable attire as less professional than other podcasts. which is frustrating as someone who enjoys the chill hippy vibe lol, i'm in chicago, so not much of a hippy vibe out here lol, but my mom has always been a legit hippy and i love it, she's also the one that got me into science as a kid, watching documentaries, and carl sagan's cosmos, and star trek, and anything that felt like exploring our reality. so for me, your whole vibe and style feels like home. but i think for some potential audiences it could feel like a lack of "professionalism", which is crazy considering the actual level of your professionalism when it comes to the things that actually matter like the content, the conversation, the intentions and goals, etc.

don't take any of this as advice though, i wouldn't be that bold or arrogant or misleading lol, i'm just offering my perspective, which may well be ignorant and unhelpful lol, and furthermore, i'm positive yall will continue evolving an intelligent balance between your own style and preferences, and that of the audiences you're looking to reach anyway! and i'm also positive that there's great success in yalls futures of science education/communication!!!

i know it must feel like a kick in the face to be having some of the highest level discussions on youtube, and even getting some of the highest level guests available, and still not having broken through yet to the size audience who should be there reciprocating your hard work and effort with their attention and interaction, but it's happening for sure, don't let any of the obstacles bring yall down, you're already doing something absolutely amazing, and you're well on your way to creating an undeniable body of work that will undoubtedly become recognizable and cherished by large amounts of grateful audience members in the not too distant future!!! love and support to both you amazing lights in this vast ocean of confusing shadows!!!! your work and effort is truly invaluable and important!!!!

[DemystifySci_Podcast]

🧠💪Listen on the go anywhere you find podcasts! Simply search "DemystifySci"

[MrVibrating]

The most readily-evident mathematical framework underpinning sensation, emotion and higher thought is obviously revealed in our affinity for music; recognising that the simplest, irreducible forms of both harmony and rhythm are factor-of-two frequency ratios - so the simplest form of rhythmic interplay, as opposed to a lone beat, is two or more beats of usually different timbre in a 2:1 (or more succinctly factor of two) relationship, and likewise in the spatial domain the simplest harmonic interval is the octave, in which we encounter the so-called equivalence paradox, such that the pattern of notes can repeat and 'pitch class' can be a thing..

The octave equivalence paradox is self-explanatory - what is 'the same' about two tones that are factors of two apart? That further is _dependent_ upon this specific spread of ratios, such that a cadence could in principle resolve successfully to any octave of the fundamental?

The stimuli aren't simply being conflated or interfering / producing some kind of distortion or channel convergence, since the effect is dependent upon their clear resolution. Rather, we're ascribing parity to information _about_ the relationship between the stimuli; the interval. The simplest interval, we're interpreting as possessing 'zero difference'.. an emergent informational zero-point in weighting relations between freq components..

The next-simplest interval is represented by all factors of three of a given fundamental, producing perfect fifths. Factors of four give octaves again, of five, fourths and of six, thirds and so on through the harmonic series..

! - we're substantiating meta-information _per se_ by weighting it entropically in terms of Shannon difference; that is, information about stimuli and the relationships between them is being encoded along an axis of convergence between thermodynamic and informational entropies; the greater that difference, the longer the temporal integration window, and thus energy costs, required to resolve it. Thus our affinity for music is based in the fact that we're processing _all_ information 'about' everything in terms of spatiotemporal modulation of factor-of-two symmetries all of the time in all modalities across the board, IOW this must be the key to the binding problem per se, not least the limbic system.

Even pyramidal cells in V1 visual cortex are representing scene depth by assigning octave-bandwidths to the cells of a projected hexagonal matrix, so we're likely applying the mechanism subliminally in all modalities.. it appears to be universally-inherent to multicellular processing.

TL;DR - an AI that perceived octaves as qualitatively equivalent in the same way as us and most animals tested would arguably qualify as the synthesis of feeling, sensation and experience. If 'qualia' are a thing, they have objective properties; their only 'irreducible' element being that informational zero-point at this default bandwidth; concisely, harmonic consonance and dissonance are but degrees of 'inequivalence' or *difference* per Shannon.. (and we should be able to replicate it in neural nets!)

[afikanyati]

I was pleasantly surprised the discussion didn’t focus on the Free Energy Principle. It was refreshing to hear new ideas from Dr. Friston. Most podcasts that have him on spend too much time trying to get him to define FEP rather than pose interesting questions for him to offer his insight into. Great work!

[MattOwens]

Suggested interview: Charles James Hall on his new book: "Beyond Relativity " regarding his proposed Hall Photon Theory.

[nolan]

$50 per spoonful from a bag of free waste?

[Thermiable]

wait. what? 9:44 "the same maths that apply to motion would apply to the way in which we use evidence to update belief structure"?

i confess that i took an instant dislike to the man, and that i deeply feel that emotion and logic are distinctly different aspects of mental space; trying to use the logical to analyze the illogical is absurd. i feel like life balances entropy (so it goes UP HILL relative to the entropic path), and that emotion balances logic; they go in literally opposite directions. i mean, i think the evidence from human behavior around money proves the fact that we are not driven by logical thought, and in what way is "following the path of least resistance" = "giving resources you need to another"; one seems like going down hill and the other, up hill, to me....

also, i had NO IDEA that such rigorous thought went into the idea that everything happens in the least energetic path possible, the shortest distance, fastest path; so he's saying that we "update our beliefs" in the same way that water flows down a hill?

quote:
Credit for the formulation of the principle of least action is commonly given to Pierre Louis Maupertuis, who felt that "Nature is thrifty in all its actions", and applied the principle broadly:

The laws of movement and of rest deduced from this principle being precisely the same as those observed in nature, we can admire the application of it to all phenomena. The movement of animals, the vegetative growth of plants ... are only its consequences; and the spectacle of the universe becomes so much the grander, so much more beautiful, the worthier of its Author, when one knows that a small number of laws, most wisely established, suffice for all movements.

— Pierre Louis Maupertuis

[margrietoregan]

you just have to ask yourself what kinds of machines what kinds of belief-based Love capable meat machines
1:06:49
can have intentions and if you just think about it all the answers you need
1:06:55
are there first of all intentions imply a capacity to act I intend to act
1:07:04
and furthermore they imply a capacity to uh to do something in the future because
1:07:10
I have not yet acted so that tells you something really important about the kinds of systems that could possibly
1:07:18
have intentions they have generative models or models not just of the future the consequences
1:07:27
of action but also their actions so they are somehow properly agents because they
1:07:35
had actions and they have the ability to countenance and to model the consequences of those actions in the
1:07:42
future that's quite remarkable with all of these machines that we are so impressed by uh in machine learning or
1:07:50
indeed that we use around the house like thermostats none of these things actually have models of the future the
1:07:56
consequences of the action they don't have intentions so the very fact you've got intentions tells you've been with a
1:08:02
very special kind of artifact or thing that has a model that is free from the
1:08:08
moment and considering the future but that even more simply things that plan
1:08:14
if you've got something that can plan that's quite a remarkable thing so to
1:08:19
have intentions means that I can indulge myself in planning and because I use the
1:08:25
word myself so now you implicitly also have this sort of notion of a minimal selfhood that's actually planning what I
1:08:32
am going to do so these are very special kinds of things and you would expect them to have a particular anatomy in a
1:08:39
particular physiology uh and a particular way of Behaving and in particular choosing who you know other
1:08:45
kinds of things of a similar sort but actually to actually engage with so yeah I think there is there is a distinction
1:08:51
between the kind of the vitalism that you're talking about um that would be present in you and your
1:09:00
pets um your parents that would probably not be in a thermostat and you could argue
1:09:07
with not even be in a virus and and of course when it comes to sort of bumblebees I don't forget what it's
1:09:13
quite clear whenever they they have that kind of vitalism or not I'm not yeah what's really interesting is some people
1:09:20
take it to an extreme I'm not sure if you're aware of the work of uh Michael Levin at Tufts but we had a wonderful
1:09:26
conversation with him where he seemed to be orienting his understanding of cellular behavioral to the cellular
1:09:34
Behavior to the level of intention as well and

[jakobsheep]

Isn't AGI mimicking emotions creating a perfect psychopath?

[ZachMeador]

this is my new favorite podcast. what a cool duo