Jeff Yoshimi
@jyoshimi
Philosopher, phenomenologist, and cognitive scientist at UC Merced. Interested in neural networks and dynamical systems theory. Builder of simbrain.net and husserl.net. Website:
From The Machine that Changed the World. Here's a time-stamped link, but the whole video is awesome. Doug Lenat's ill-fated efforts to codify commonsense knowledge with a database of propositions is especially fascinating. youtu.be/enWWlx7-t0k?...
Hubert Dreyfus on common sense knowledge circa 1992. I think it's held up well!
This screenshot shows a bit of the CNN visualization. Click through tensor slices to see how different representations of inputs are learned. If you hover over a pixel, upstream and downstream patches of influence are highlighted. Visualize strides, padding, receptive fields, etc.
Simbrain 4.1 is out! A major update adding support for convolutional networks, genome visualization, speech synthesis, a fun new “field of consciousness” plot, and more. Shown here: Sejnowski and Rosenberg’s model of reading aloud, which helped spark the connectionist revolution of the 1980s.
I have a co-authored chapter that does just this! Moreover, even if not quite what you want, it's easy to edit and change to your liking, and I'd be happy to work with you to create a fined version. downloads.jeffyoshimi.net/NeuralNetwor...
How is this blocky, doodly painting supposed to conjure up Benjamin's world-historical ruminations? I like abstract art but this one bothers me somehow, and certainly does not feel up to the task Benjamin sets for it. Any art snobs out there willing to set me straight?
We’ve added a graphical interface to the neuroevolution framework in #Simbrain. Watch as genes mutate and see how this impacts fitness. In this example, two networks controlling two agents are evolved and fitness corresponds to how much they’ve eaten.
I had my best day ever playing an RPG (Dungeon Crawl Classics) yesterday. I used lotus stare to control a spellcasting creature who led us to the final boss and covered gnolls in spider webs, and in three battles I used word of command to engage enemies against one another.
A pigeon goes in, a classification comes out! #Simbrain now supports deep networks and CNNs. As activity propagates through the network, progressively more abstract features are formed, culminating in a representation sufficient for classification in the final layer.
I was recently on the podcast “Reality Gone Mental”: youtu.be/vk1K9x-qGOU It was a refreshingly exploratory discussion. My take on the mind-body problem is that it can’t be solved so we should focus on the more tractable project of neurophenomenology... 1/
For more on Føllesdal’s life and work and the school of thought he helped give rise to, see this paper jeffyoshimi.net/downloads/Ca... Some relevant passages attached. The last two are quotes from Føllesdal himself, which highlight his willingness to depart from Husserl when necessary.
Dagfinn Føllesdal passed last week at the age of 93. He was my philosophical grandfather (my advisor's advisor), and was known as a founder of the “California School” of phenomenology. Here he is with his students David Smith and Ron McIntyre, as well as David’s students: Phil Walsh and myself.
I’m thrilled to announce the release of #Simbrain 4, which I have been working on for over 10 years! For the last five of those years I have met almost every weeknight with Yulin Li and we have pair-programmed our way through the entire app, adding hundreds of new features.
Consciousness is not a sequence of isolated qualia, but a rich and structured field, with many interacting parts. These “parts” are like members of a symphony, whose contributions are distinctive but interwoven.
I was recently on Tonya Kubo’s podcast, discussing the history of citizen science, what makes games so powerful and engaging, and how playing and building new scientific discovery games can be a way to channel grief towards collective action.
Heidegger dedicated _Being and Time_ to Husserl “in friendship and admiration” and presented Husserl with a bound copy around the time of his 68th birthday. As Husserl read the book he became increasingly aware of the distance between them. 1/3
Clark Hull is often grouped with the behaviorists or neo-behaviorists. He emphasized stimulus and response, but also allowed for a lot of stuff going on between the two. It’s like neural nets before there were neural nets. Check out these pictures, which are from his writings in the 1930s and 40s.
Braitenberg vehicles are known for how complicated their behaviors can be, despite how simple they are: a few sensors, a motor, and some weighted connections. Put two vehicles together and an even greater wealth of behaviors can occur.
Neural nets have clearly shown their power in the intervening decades, but are we any closer to a "definitive assessment of their power and validity"? FWIW I think Smolensky's approach has held up. The mind ain't no LLM, but cognition is still well-approximated by continuous subsymbolic dynamics.
A talk I gave on mathematical phenomenology last week is now up on youtube. Topics include whether phenomenology can be "mathematized", Husserl's time diagrams, and efforts to formalize the horizon structure. youtu.be/ITMLgz7OLrA
We’ve recently updated our collaborative open-access book, “Neural Networks in Cognitive Science”, adding a few new authors, chapters, and lots of content. downloads.jeffyoshimi.net/NeuralNetwor...
What do you do when you come across a symbol you can’t pronounce? Suppose it’s something you have to deal with and can’t just ignore. I find it remarkably unsettling.
A much earlier entry in the Chronik, from the other bookend of Husserl's professional life (the student years, 1884/6), confirms Gregor's point about Brentano introducing Bolzano to Husserl. Interestingly, Osborn is again quoted, this time saying Brentano familiarized Husserl with Bolzano's work.
Here is the source of the story about Husserl discovering Bolzano in a second-hand bookstore. It's from late in Husserl's life, 1935. We have to rely on Spiegelberg recounting something Osborn said, so there are a few sources of potential error
I already reposted twice about this yesterday, but allow me one more post, this time with my own reading notes! So, last week I was on @trsam97.bsky.social 's podcast talking about Merleau-Ponty with @robinmuller.bsky.social ... youtu.be/YPmuFqUwSLw