iiinova — where your digital self keeps evolving

Darius Valevicius @dariusliutas.bsky.social · 27/07/2026

As sometime who works in precisely that intersection, I can confirm that it's a lonely world out there :')

Darius Valevicius @dariusliutas.bsky.social · 18/07/2026

I can't find the source, but I remember a quote from a paper from the 70s or some such that went something like "the purpose of conferences is for morale, and to feel like something is happening when nothing is happening".

081

Darius Valevicius @dariusliutas.bsky.social · 17/07/2026

In a sixplex with an interior stairwell!

010

Darius Valevicius @dariusliutas.bsky.social · 17/07/2026

My roommate has told me that my cat always starts meowing at the door when I'm about to come back and never when it's another person. I hear him from the moment I open the door to the foyer 🫪

210

Darius Valevicius @dariusliutas.bsky.social · 02/06/2026

P.S. Make sure to check the extensive supporting document for the full methods and results - lots of design decisions documented within which can help with using the method in your own experiments :)

130

Darius Valevicius @dariusliutas.bsky.social · 02/06/2026

Thank you to my supervisor @vincenttd.bsky.social and colleague Celine Haddad, as well as our collaborators @neuroaure.bsky.social and Gurgen Soghoyan...

120

Darius Valevicius @dariusliutas.bsky.social · 02/06/2026

Coming soon - we've been using this method with fMRI to try to figure out what emotion-related BOLD signals are directly representing subjective fear (and emotion generally?) and which are secondary correlates. I'll be presenting the results of that at #ASSC this summer! ...

121

Darius Valevicius @dariusliutas.bsky.social · 02/06/2026

... zero-in on the preferred stimuli of different emotion-related variables. We find some interesting stuff with regards to the concordance between self-reported fear and physiological measures (i.e. skin conductance responses)...

110

Darius Valevicius @dariusliutas.bsky.social · 02/06/2026

Excited to share my first doctoral paper which was just published at Psychiatry and Clinical Neurosciences! doi.org/10.1111/pcn.... In this research letter, we showcase our take on the "Synthetic Image Evolution" method, in which we use genAI and evolutionary algorithms to... #neuroskyence

doi.org

Synthetic image evolution for affective science

Click on the article title to read more.

274

Darius Valevicius @dariusliutas.bsky.social · 21/05/2026

You can be a monist physicalist/materialist and still puzzle over the hard problem of consciousness, no? This article seems to conflate the hard problem with dualism.

030

Darius Valevicius @dariusliutas.bsky.social · 13/05/2026

Wrote a little microblog on some thoughts on interoception and emotion, curious to hear what y'all think and might develop it into a longer blog post if it sparks any interest :) @nicolecrust.bsky.social @micahgallen.com #neuroskyence

162

Darius Valevicius @dariusliutas.bsky.social · 13/05/2026

Finally, how can we best explain the range of human values that go beyond material self-interest, e.g. sociality, spirituality, morality, aesthetics, etc.? Is an interoception-only account sufficient?

100

Darius Valevicius @dariusliutas.bsky.social · 13/05/2026

So for me an interesting question is: What is the architecture of the value function? How are both (or either) interoceptive and exteroceptive senses linked to pleasure and pain via evolution? What is innate vs what are learned contingencies, predictions, or elaborations?

100

Darius Valevicius @dariusliutas.bsky.social · 13/05/2026

Without positing care or love as basic emotional systems, it becomes very difficult to explain cases of self-sacrifice. There is a bit of a parallel for me between the interoceptive RL view and the "rational actor" theory of economics, where material self-interest is the only basis for behaviour.

110

Darius Valevicius @dariusliutas.bsky.social · 13/05/2026

Either this is the basis of a non-interoceptive emotional system, or it is a contingency learned between touch and homeostasis (i.e. touch induces a low-stress state, and that physiological state is what is experienced as pleasant and gives rise to bonding and separation distress). Could be either?

100

Darius Valevicius @dariusliutas.bsky.social · 13/05/2026

We seem to be born with a need to experience a particular kind of touch, which likely evolved to keep us sticking together as that kept us alive. So a particular exteroceptive sensation is tied to a particular kind of pleasure and pain, and this is so at birth (without touch newborns sometimes die).

100

Darius Valevicius @dariusliutas.bsky.social · 13/05/2026

Another counter-proposal comes from Panksepp's framework, where in addition to "sensory affects" (more like interoceptive theories), the emotions are tracking variables that would be a stretch to label physiological. For example, separation distress is an early response to lack of touch in newborns.

100

Darius Valevicius @dariusliutas.bsky.social · 13/05/2026

It could be that the coupling that develops between predictions of future homeostasis and valence progressively decouples emotion from the current physiological state? Which makes the hypothesis hard to falsify, but...

100

Darius Valevicius @dariusliutas.bsky.social · 13/05/2026

One problem is that there is very little in the way of correlations between physiology or interoception and emotion, and the behavioural evidence that the "allostatic-interoceptive" network of Feldman Barrett et al. is really about interoception seems scant to me.

100

Darius Valevicius @dariusliutas.bsky.social · 13/05/2026

These views posit that an organism's goal is to maintain its physiological parameters in a viable range, and that contingencies and predictions are learned that link up our perception of the world to this internal cost function. Thus, things become valenced based on how they will impact the body.

100

Darius Valevicius @dariusliutas.bsky.social · 13/05/2026

Is emotion interoceptive? While I think the role of emotions is basically a kind of conscious representation of a cost function, in line with recent frameworks like "homeostatic" or "interoceptive" reinforcement learning, I'm not convinced that the basis is really in the internal sense of the body.

272

Darius Valevicius @dariusliutas.bsky.social · 13/05/2026

Since incontrovertible evidence is basically impossible, the difference is in the priors. A cat is basically 99.9% similar to me, an ostensibly conscious being, while a linear regression on steroids running on rocks, plastic, and metal is much different!

010

Darius Valevicius @dariusliutas.bsky.social · 12/05/2026

The less I use it, the more I find myself much more affected by media that deals with the grief of mortality, which is at once piercingly beautiful in its own way, and to me feels lonely since it doesn't seem to resonate as much with others

000

Darius Valevicius @dariusliutas.bsky.social · 20/04/2026

Featuring some accordion, (shaky) violin, and some Greek lute 🎵🎶 And sad vibes

000

Darius Valevicius @dariusliutas.bsky.social · 20/04/2026

I set up a Bandcamp to host a bit of music I wrote for a game jam this past weekend, hopefully more such projects to come :) dariusvalevicius.bandcamp.com/album/fear-n...

110

Darius Valevicius @dariusliutas.bsky.social · 17/04/2026

Given that the task outcomes also don't correlate with ADHD, could this be an instance of Wittgenstein's Ruler? I.e. Does it tell us more about the task's limitations than people's real-world attentional capacity?

100

Darius Valevicius @dariusliutas.bsky.social · 27/02/2026

It also seems like the only thing I've encountered that satisfactorily operationalizes/grounds @neddo.bsky.social's and @anilseth.bsky.social's claims that consciousness might have substrate dependence in a concrete argument (the argument that only fields can physically integrate information)

010

Darius Valevicius @dariusliutas.bsky.social · 27/02/2026

Still waiting to hit a wall with CEMI field theory. I've sent it to two physics grad student friends who say the physics is sound - no evaluations yet from any computational neuroscientists! Surely someone can poke a hole in this for me? #neuroskyence

130

Darius Valevicius @dariusliutas.bsky.social · 19/02/2026

Very different from LLMs that are simply fed context/memory back in as external input each time they're asked to generate a token. That's cheating!

010

Darius Valevicius @dariusliutas.bsky.social · 19/02/2026

I've been interested in models of working memory that use reservoir-like populations of recurrent neurons in the PFC that pool information and keep the echoes of information alive over time using feedback loops ("echo state machines")...

110

Darius Valevicius @dariusliutas.bsky.social · 19/02/2026

(context integration beyond one or two neighbouring words*)

110

Darius Valevicius @dariusliutas.bsky.social · 19/02/2026

3, no transformer-like circuits found in the brain AFAIK? Context integration ("self-attention") of language semantics in the brain seems to be done by conscious working memory mechanisms according to work by Dehaene, and isn't done in an automatic/feedforward way. youtu.be/V1OaDKuffBM

youtu.be

Qu’est-ce que la conscience, et quels sont ses mécanismes cérébraux ? (2) - S. Dehaene (2025-2026)

YouTube video by Sciences de la vie - Collège de France

100

Darius Valevicius @dariusliutas.bsky.social · 16/02/2026

But in my consciousness, I have access to both the low level (uncompressed) representations and the concepts at the same time. My awareness isn't swimming in a dark world of ideas except when I close my eyes (and ears and...). How is all this information bound together? ...

100

Darius Valevicius @dariusliutas.bsky.social · 16/02/2026

The proposed solution to the binding problem is really interesting and touches on something that's been bugging me about other theories like HOT or GWT. Our current best models compress sensory representations and feed them into a latent space that can perform computations on concepts...

100

Darius Valevicius @dariusliutas.bsky.social · 16/02/2026

Does anyone fuck with the CEMI field theory of consciousness? Compatible with recent work on ephaptic coupling championed by @earlkmiller.bsky.social and surprisingly parsimonious... Briefly, the unconscious mind == classic neural computation, concious mind == EM field computation

120

Darius Valevicius @dariusliutas.bsky.social · 27/01/2026

My greebles aren't behaving so I've been thinking about my backup plan to open a bakery or become a blacksmith or something

000

Darius Valevicius @dariusliutas.bsky.social · 22/01/2026

Reposting this earlier edit of another film but more legible. Same vibes

100

Darius Valevicius @dariusliutas.bsky.social · 22/01/2026

I had a transcendent experience after watching Resurrection (2025) in the cinema. Standing on the sidewalk chugging cigs for 25 minutes, staring wide-eyed down the street waiting to come back to reality.

100

Darius Valevicius @dariusliutas.bsky.social · 16/01/2026

I think we made a mistake when we stopped being little rat creatures, and a second mistake when we accepted modernity. Grooming, cuddling, co-sleeping with your fellow rats, all lost to time.

140

Darius Valevicius @dariusliutas.bsky.social · 13/01/2026

... Especially in young humans or young animals, who haven't had time to develop sophisticated statistical world models. But maybe I'm missing part of the picture. Any thoughts? @nicolecrust.bsky.social @jtheriault.bsky.social

220

Darius Valevicius @dariusliutas.bsky.social · 13/01/2026

This came up in a paper (Yin 2014) on basal ganglia control of movement. It reminds me of a critique I had of Feldman-Barrett's predictive processing theory of affect - it seems computationally intractable to predict, with a forward model, the metabolic consequences of everything that happens...

220

Darius Valevicius @dariusliutas.bsky.social · 12/01/2026

"Yet modern man feels uneasy and more and more bewildered. He works and strives, but he is dimly aware of a sense of futility with regard to his activities. [...] While becoming the master of nature, he has become the slave of the machine which his own hands built." (E. Fromm 1947)

000

Darius Valevicius @dariusliutas.bsky.social · 12/01/2026

Long ago, the four dynamical regimes lived in harmony. Then everything changed when

000

Darius Valevicius @dariusliutas.bsky.social · 12/01/2026

To people who are afraid of rats/mice - would it help if I told you they have human emotions?

000

Darius Valevicius @dariusliutas.bsky.social · 11/01/2026

That said I think I might agree with Yoshua Bengio's take (and even going back to the paperclip thought experiment) that giving AI/AGI drives is something that we should actually avoid at all costs, lol

130