The Cognitive Design of Tools of Thought (2014) [pdf] | Hacker News Reader

The Cognitive Design of Tools of Thought (2014) [pdf]

gjgreenberg.bol.ucla.edu

185 pointsandsoitis21 comments

"The Extended Mind" by Andrew Clark and David Chalmers, is a much more interesting (to me) examination of the categorical blurring between the mind and externalized tools of thought.

https://consc.net/papers/extended.html

This idea is also one of the many interesting facets of my favorite novel, "Jonathan Strange and Mr. Norrel".

https://joeldueck.com/mind-palaces.html

A common theme is that tools for thinking allow you to restimulate your brain with its own output. Essentially creating a full outer loop from your sensory inputs to your motor outputs.

I'm speculating now: There are probably smaller loops within the brain that allow you to ruminate on a topic to think through it, but those loops don't give you access to as many neurons or have the same fidelity between iterations. The alternative to this sort of self-stimulating recursion, is that everything we reason about can be done in one pass from sensor to motor, which I find unlikely.

The idea of applying multiple passes of the same networks using memory to hand off between iterations also explains why memory and other aspects of intelligence are correlated in humans, when they needn't be. A machine has FLOPS independent of it's memory capacity for example.

Isn't that close to what instruct-tuned LLMs do when they are adviced to first think out aloud before jumping to answering? They are generating their own output to their scratchpad, the attended context, to support their thinking.

Coincidentally, the author, Dr. Barbara Tversky [0], was married to Amos Tversky, of Kahneman and Tversky fame — their decades-long collaboration was the subject of one of Michael Lewis's books [1]. After Amos Tversky's early death, Barbara Tversky and the also-widowed Daniel Kahneman became partners until Kahneman's own death just a few weeks ago.

[0] https://en.wikipedia.org/wiki/Barbara_Tversky

[1] https://en.wikipedia.org/wiki/The_Undoing_Project

Wow, that's fascinating. I didn't know Kahneman and Tversky were related through Barbara as well.

This topic has been researched and discussed extensively in the embodied cognition branch of cognitive science. It's mentioned as a central aspect of cognition in Margaret Wilson's landmark paper " Six views of embodied cognition" (https://pubmed.ncbi.nlm.nih.gov/12613670/) which came out 12 years before this one. I'm surprised to not see much of the extensive research that has been on this idea before cited here.

Direct PDF link: https://link.springer.com/content/pdf/10.3758/BF03196322.pdf

If you're interested in this topic but want more, I recommend "Cognition in the wild": https://mitpress.mit.edu/9780262082310/

What a waste of time. Who provides funding for this? Also mildly annoying to use "isomorphism" outside of mathematics to describe a relation or analogy between things. You can just say relation or analogy.

Relying on the technical mystique of precicely defined jargon in a context where it doesn't apply should indicate that there is a lack of substance in the ideas being presented, which is why the author feels compelled to obfuscate it.

Here's an analysis of similar ideas, but in the context of business strategy: https://onlinelibrary.wiley.com/doi/10.1002/smj.3613

Looking at Figure 5, I'm reminded of a distinction I like to make between "cookbook" instruction and "textbook" instruction. The former is like the top of the figure: everything is fully in context, but also everything remains implicit, and it's up to the reader to suss out* the principles given a bunch of examples. The latter is like the bottom of the figure: major things are pulled out of context, explicitly so that their principles can be communicated, noninductively and clearly.

* That which thy fathers have bequeathed to thee, earn it anew if thou wouldst possess it. — JWvG (as quoted by ISR)

> When thought overwhelms the mind, the mind puts it into the world, notably in diagrams and gestures

And in song and words.

> Some ancient outpourings of thought ...

And oral histories.

So, using media to help with cognition. Odd that they don't mention oral media.

For some reason, oral media are not as good at this as the written word and abstract drawing. James Gleick (The Information) speculates that it has to do with being able to zoom back and forth between parts of sentences and paragraphs. It offers an explicit way of structuring thought.

I think of rhetorical devices as a way of formalising oral discourse, providing signposts that aid the listener "parse out" a more-complex-than-1D structure, aiding "zooming back and forth between parts of sentences and [whatever the oral equivalent of a paragraph may be]" via explicit additions.

(I wonder if the different temporal aspects of delivery also play into other aspects of style? For instance, I translated an italian art critic once whose prose style was heavily pyramidal: every so often, at the beginning of a paragraph, he'd introduce a crisp new thought, and his language would steadily get more elaborate and flowery after that, until I often was struggling, but just before he seemed about to leave the plane of actual semantic meaning behind, he cycled back to the next crisp thought. My hazy understanding of latin oratory is the opposite: it was supposed to be all flowery at the outset, and steadily tighten to conclude crisply.

If this be a true distinction, it'd make sense in the light of media consumption: more people read a hed than lede, more the lede than anything after, but presumably orators attract listeners, so more people hear the conclusion than prepatory statements, and more the prepatory than the prefatory?)

Of possible interest

https://www.rheingold.com/texts/tft/

There's a few cartoons going around where they describe the effect of various drugs using graphical patterns, facial expressions, etc.

That's a rich vein for development. It could be generalized and optimized.

Because states of consciousness are so central to our ... everything.

And it's a subject that's difficult to discuss. I imagine well-designed models rendered with nice animations.

We could discuss drugs, dreams, focus, meditation, fiction...

In a published book I've seen a nice set of facial expressions, with a small (half dozen?) set of "eigenemotions", then the matrix of combinations of the eigenemotions to express more nuanced emotions.

Anyone know of a publicly-accessible image in this genre?

(Laumer's Corps Diplomatique Terrestrienne[sic] has gone so far as to give each facial expression an official alphanumeric code, but I doubt he bothered to substructure, or even hold consistent, the coding scheme)

Will Eisner's, Comics and Sequential Art, book is great for graphic storytelling and the history of the artform.

Is culture also a tool for thought?

Defo. (to the point that people who are only aware of their mother culture don't even think of it as a tool; it's just what they always use, whether they're hammering nails with it — or screws)

How will this change over time as we get to further iterations of Neuralink? Or even Apple Vision?

Reply on news.ycombinator.com