Why do we fall prey of illusions such as the one below?
The answer lies in how our brain is wired
(thread, 1/N)
2/ Let's look at a fragment of the previous image.
It looks coherent and could plausibly exist.
3/ Also the other two fragments, when examined one by one, look plausible.
4/ The thing is, our brain does not examine the image as a whole (even though we have the impression it does so).
If it did, it would notice the logical impossibility.
5/ Instead, our brain considers the image as a patchwork.
Coherence is evaluated at this level, one piece of the patchwork at a time. If each looks coherent, our brain will *intuitively* think that the image is okay.
Hence the illusion.
6/ Of course, we have powerful analytical capabilities. If we focus, we can examine the image as a whole and notice the logical incoherences.
Analytically, it seems wrong.
Intuitively, it seems right.
Hence the awkward feeling from the illusion.
7/ Our brain does not evaluate large pieces of information as a whole. At least intuitively, it breaks them down in pieces and evaluates each separately, leading to some of the quirks in our functioning.
I call this the Distributed Brain Framework.
8/ It's an important topic, for it explain why we sometimes decide something but do something else. Or why we hallucinate. Or why we confabulate explanations for our behavior.
However, it's too complex to explain in a Twitter thread, so I decided to organize a free talk about it
• • •
Missing some Tweet in this thread? You can try to
force a refresh
“Create a UserPromptSubmit hook (global settings). Script echoes: If 8+ tool calls, append one optimization hint (reusable skill, memory pattern, or workflow fix). One sentence. Skip if exploratory.”"
2) Skills audit
"Create a skill that lists all my installed skills (project & global level) with their line counts. Then ask the user which to review for improvement opportunities (conciseness, clarity, overlapping scopes, token efficiency).”
3) Claude audit
“Create a skill that reads all CLAUDE .md files and checks for: redundant instructions, verbose phrasing, and content that could move to memory. Present findings and ask if the user wants to implement them.”
Highlights from today’s Jeff Bezos’ talk in Turin 🇮🇹:
“Advice to young people: go work to a company where you can learn best practices”
I fully agree: it should also apply to politicians, educators, and other high-leverage roles.
1/N
“You can be an entrepreneur within a company; good companies don’t eject mavericks but empower them.”
I add: it’s so important to select a great first job and first boss; it’s sad it’s mostly left to chance, esp. comparing how much time is spent studying and how little interviewing.
2/N
We interviewed @linaashar, founder of Dreamtime Learning, who has very interesting thoughts about education.
Some of my favorite quotes:
“I keep teaching kids about their brains and their behavior in every session. Because if kids can master their brains, their thoughts, their actions, and therefore their behaviors, they're going to be successful. That's a given. But if they master only what is calculus, or what this is and what that is, even though they may get an A+, success is not a given. Because you can master content, but if you have to master yourself, you're lost.”
(link at the bottom; 1/7)
“We do not [as society] design the education system or the learning sessions in the way their brain actually works.”
2/7
“If their whole school time is spent on learning the core curriculum, where is the time for kids to specialize? Where do they get those 10,000 hours that they need to become a specialist? So you have to free up time in the child's day for them to become highly specialized.”
3/7
I recently got a small grant (courtesy of Kanro, Vitalik Buterin's foundation) to produce some educational materials regarding the pandemic response.
These 10 one-pagers are the first batch of educational materials.
Any feedback?
1/10
Some more background about the one-pagers. They are meant for people who are already onboard with the need to properly react to an eventual future pandemic but don't have the vocabulary or examples to explain to others what they can do and why.
2/10
A simple model to understand indoor infection risk