https://en.wikipedia.org/wiki/Public_Opinion
Frederic Bartlett (1932) defines schemas as memory structures that pre-shape perception and recall:
https://en.wikipedia.org/wiki/Schema_(psychology)
Jean Piaget explains schema updating via assimilation/accommodation when evidence conflicts with the map:
https://en.wikipedia.org/wiki/Assimilation_(psychology)
Edward Tolman introduces cognitive maps, making "map" literal in psychology:
https://en.wikipedia.org/wiki/Cognitive_map
Marvin Minsky formalizes frames as slot-filled expectations that speed inference but can blind you to anomalies:
https://en.wikipedia.org/wiki/Frame_(artificial_intelligence...
voidhorse: "mental model" vs "theory" is a real distinction in the literature. Kenneth Craik frames small-scale models as internal simulations for reasoning, not public theories:
https://en.wikipedia.org/wiki/Kenneth_Craik
Philip Johnson-Laird formalizes mental models as internal simulations used for inference and prediction:
https://en.wikipedia.org/wiki/Philip_Johnson-Laird
andsoitis: "informal, simplified, personal" models are exactly why systematic errors show up. Daniel Kahneman and Amos Tversky document heuristics and biases when internal maps are over-trusted:
https://en.wikipedia.org/wiki/Heuristics_in_judgment_and_dec...
Repair loop: Seymour Papert's microworlds provide controlled sandboxes for testing and revising models:
https://en.wikipedia.org/wiki/Constructionism_(learning_theo...
Gary Drescher gives a schema mechanism for incremental action/outcome updates that rebuild the map from experience:
https://mitpress.mit.edu/9780262517089/made-up-minds/
If you want to see Drescher operationalized, MOOLLM turns the schema mechanism into working skills. Schema Mechanism is the causal core, Schema Factory adds a deterministic toolchain and context bundles for LLM reasoning, and Play-Learn-Lift is the governance loop that maps ACT/OBSERVE/ATTRIBUTE/SPIN OFF into audited upgrades. This is GOFAI made practical with LLMs filling the old gaps in grounding and explanation.
Drescher's Schema Mechanism as Anthropic Skill:
https://github.com/SimHacker/moollm/blob/main/skills/schema-...
Drescher's Schema Factory as Anthropic Skill:
https://github.com/SimHacker/moollm/blob/main/skills/schema-...
Play=>Learn=>Lift methodology as Anthropic Skill:
https://github.com/SimHacker/moollm/blob/main/skills/play-le...
Here is the exact kind of thing we are talking about -- the YAML Jazz schema examples are live, readable schemas-by-example with causal context, semantic comments, evidence counts, side effects, and marginal attribution notes, including a practical devops edgebox/ingest cluster and a Zork/MUD "learn by dying" cluster so you can see the mechanism at work in real data:
https://github.com/SimHacker/moollm/blob/main/skills/schema-...
# YAML Jazz schema examples (comments are semantic)
#
# These are schemas-by-example: minimal structure, rich intent.
# Follow canon schema rules where possible, but annotate as needed.
# Ad hoc fields and side-notes are allowed for partially jelled ideas.
https://github.com/SimHacker/moollm/blob/main/examples/adven...
Finally, if you want the deeper connections tour written specifically for this thread -- the big-picture synthesis that ties Papert, Minsky, Drescher, Play-Learn-Lift, and live microworlds into one operational map -- dive here:
https://github.com/SimHacker/moollm/blob/main/designs/CONNEC...
There's an argument to be made that it is useful to distinguish between mental models and theories.
If a theory is a structured, formal explanation of phenomena, grounded in evidence, logic, and often mathematics that is meant to be shared, tested, and and falsified, a mental model is more of an internal representation of how something works, often informal, simplified, personal, and built to help you reason, predict, and decide.
I find both tools useful, but different.
Always a good read
Mental Models: The Best Way to Make Intelligent Decisions - https://news.ycombinator.com/item?id=24527003 - Sept 2020 (35 comments)
Mental Models: The Best Way to Make Intelligent Decisions (113 Models Explained) - https://news.ycombinator.com/item?id=17121145 - May 2018 (36 comments)
You can use the Wayback Machine to read the version that was originally discussed.
Yeah yeah, like someone is doing charity here.
Will always be grateful to Shane for that!
Shane's mental models books are packed with a lot of random/disparate domains/insights -- He's a good aggregator there.
Thinking in Systems by Meadows.
Really, once you go down the rabbit hole, you find new threads to pull. That's kind of the fun of it
But what I really wanted to say, this reminds me of Scott E Page’s Coursera course on Model Thinking, and a book: “The Model Thinker What You Need to Know to Make Data Work for You” also from 2018.
A transcript of Charlie's speech is still up https://fs.blog/great-talks/a-lesson-on-worldly-wisdom/
I guess Shane Parrish is trying to carry the torch on now that Charlie has passed.
That said, I am on mobile…?