Sambourne House
Sambourne House in London - a well-preserved Victorian middle class home, decorated by the Punch cartoonist Linley Sambourne - is a dream. I didn't want to leave
Typical of the waking self to think sleep is strange and wonder why we need it. Unconscious sleep is a return to the ordinary way of things. Wakefulness is just a neat trick for a certain kind of unconsciousness to sustain itself. How odd that we spend 2/3 of our lives awake aeon.co/essays/sleep...
This web version of Werner's Nomenclature of Colours by @rougeux.bsky.social is glorious. This is emblematic of what the internet should be for - edifying and beautiful
A lot of good sense in this one
There's an important difference between a tool having a formal structure that we learn to use skilfully and a tool just being a kind of learned skill. Is our understanding of a mathematical operation an understanding of how to apply a rule or just a behavioural regularity that we look at abstractly?
Some refreshing conceptual clarity from @timbayne.bsky.social on ways of defining consciousness, though I think there's more of a break between his view and the mainstream one than this article explores
Or (just maybe) what you think you know so far contains a misunderstanding and your ‘hard problem’ is a warning of an unseen contradiction.
I would sooner assume that I’m vainly trying to explain something I’ve misdescribed than posit a parallel universe of reified sensations or sentient electrons.
There's a terrible hubris in the label ‘the hard problem of consciousness’. It says: we have understood everything so well so far that this thing we cannot understand must indicate a mystery inherent in the fabric of the universe-
Best title for a paper I've seen for a while: 'Is the dynamical quantum Cheshire cat detectable?' arxiv.org/abs/2204.03374
Gauss: easy winner for best signature?
The behavioural data is well worth thinking about carefully but I don't think the headline result is technically demonstrated
Interesting results but the paper doesn't define compositionality robustly enough. Condition (iii) - the meaning of the sum is "derived" from the parts - is a synonymous description, not a definition. Then measuring it as relative Euclidean distance in a semantic space is a weak proxy at best
For some, this is a contradiction: to say something anthropomorphic about a model is the same thing as making a claim about its 'real' intelligence. But rather than accept the equivalence and deny the language (which is futile), we should point out the conceptual error
I don't think current AI models are human-like but I also don't think anthropomorphic descriptions are problematic. Anthropomorphism is core to perception and language (it's the intentional stance) - better just to dissociate it from psychology, as when describing reasoning in e.g. insect behaviour
For some, it informs their ideas about moral responsibility and therefore their political opinions and therefore their voting behaviour and so on. But what's in scope of what you mean by "how you live"?
