Join 3,557 readers in helping fund MetaFilter (Hide)


Braitenberg vehicles: How to build a brain
January 17, 2010 9:10 PM   Subscribe

Valentino Braitenberg's 1984 book, Vehicles: Experiments in Synthetic Psychology was a seminal work for its discussion of how one might design a system (biological or otherwise) in order to generate behavior like that seen in beings with brains. He embarks on a series of thought experiments in which he creates thirteen "vehicles" through simple components that (arguably) display intelligent behavior, evolving in a Darwinian fashion to demonstrate what appears to be high-level cognition.

In the past 26 years, with the expansion of internet access and capabilities around artificial intelligence, a number of people have attempted to make Braitenberg's vision into a reality. Beautiful artwork has been inspired by the plans for the vehicles. Now, online simulators allow laypeople to play with the vehicles. You could even build one yourself!

Although fascinating at face value, the true depth of Braitenberg's work (as is common with artificial intelligence) may arguably lie in its philosophical and ethical implications. What does this mean for how we conceptualize mental illness and other dysfunctional behavior? How does it affect what we define as free will and intelligent thought?
posted by emilyd22222 (16 comments total) 35 users marked this as a favorite

 
Craig Reynolds is one of the biggest contributors along these lines. It is interesting how simply trying to make something look lively ends up teaching us things about life. Analysis by synthesis.
posted by StickyCarpet at 9:34 PM on January 17, 2010


Woot! We played with Braitenberg vehicles in grad school. Love it!
posted by leotrotsky at 9:35 PM on January 17, 2010


A Linux Braitenberg machine simulator, by a guy who makes very funny web pages for his applications.

"DEDICATED TO THE PRINCIPLE THAT ANYTHING CAN BE RUTHLESSLY OVERENGINEERED"
posted by idiopath at 9:56 PM on January 17, 2010


By the way, all of my comments in this thread are licensed under the Open Profanity License.
posted by idiopath at 9:58 PM on January 17, 2010


Interesting.
posted by delmoi at 10:39 PM on January 17, 2010


Hell yeah--one of my favorite books of all time. These online simulators are awesome.

And idiopath beat me to it, but Prof. Kohler is awesome as well. Be sure to take his classes if you're ever at UCLA...
posted by equalpants at 10:40 PM on January 17, 2010


Living organisms don't evolve using "intellegence". Evolution is like trial and error; mutations that work well let the organism procreate, mutations that don't work well make it harder for the organism to procreate.

The only people that use "intelligence" in regards to evolution are the people who are using fancy-schmancy marketing to teach creationism.

I think what these cars are doing is "learning" not evolving like the Darwin sense of the word.
posted by hal_c_on at 11:02 PM on January 17, 2010


''Gawrsh!'' /Goofy
posted by y2karl at 11:17 PM on January 17, 2010


I think what these cars are doing is "learning" not evolving like the Darwin sense of the word.

They do both, actually. In some of the thought experiments, the vehicles have learning mechanisms, but in others, there's a Darwinian component--the vehicles are driving around on a big table, which they can drive off the edge of and die; also, a machinist grabs living vehicles at random and makes copies of them (possibly making errors, of course).

Braitenberg is actually very careful not to make any grandiose claims about "intelligence" or whatever--he's pretty clear that the purpose of the book is just to illustrate how simple mechanisms can give rise to behavior that's difficult to analyze.

Admittedly, it's pretty tempting to generalize from that to "human behavior is probably simpler than we think", and there's definitely an undercurrent of that running through the book, but he's careful not to make explicit claims about such things. A large portion of the book is devoted to real-life biological examples--he points out some of the mechanisms that inspired the vehicles, and shows how simplified the vehicles are compared to the real thing. To me, this is what puts this book in a different league from Gödel, Escher, Bach and similar stuff: there's the same sense of wonder, but the caution and rigor keep everything grounded.
posted by equalpants at 12:23 AM on January 18, 2010 [1 favorite]


Seems like a novel way to approach behaviorism, very interesting.
posted by afu at 1:24 AM on January 18, 2010


The first pages I ever put on the web were about my Braitenberg vehicle simulator and I recently made the Lisp and Java source code of the simulator available (I haven't documented the code yet, let me know if you have trouble using it).

You can watch videos of some test runs from the simulator, including a demonstration of communication between vehicles where there are predator and prey vehicles, and prey vehicles give an alarm cry when being chased.

My simulator is based on the A. K. Dewdney Computer Recreations article about Braitenberg.
posted by jjwiseman at 7:31 AM on January 18, 2010


Aren't these "vehicles" actually dumbed-down expressions of human intelligence and behavior?
posted by KokuRyu at 9:10 AM on January 18, 2010


Wow. When I was about 13 years old, I skipped school a LOT... And would head to the University Library to check out things that REALLY interested me, and this was one of those books. I had originally read about it in Scientific American (my father was an avid reader), and when I read about this book, I thought it was the coolest thing EVER...

My local (public) library didn't have this book, and the only place I could find it was the university, so I headed there and *wished*, oh how I wished, that I could build such beasts. But then I'd have to tell my parents that I'd skipped school, you see, so such plans never came to fruition.

Thanks for bringing this back to my world. :)
posted by Vamier at 10:15 AM on January 18, 2010


Aren't these "vehicles" actually dumbed-down expressions of human intelligence and behavior?

They are oversimplifications, yes, and they are also interpreted from a human lens- we call the behavior of these machines "love" or "fear" just because it looks familiar to us. The point, as I see it, is that it's possible to make these behaviors that, in humans, appear very complex, with very simple components. This has implications for how we move forward in understanding human cognition. Twenty-six years later, there is still much we don't understand.
posted by emilyd22222 at 10:40 AM on January 18, 2010


Like others, I was fascinated by Braitenberg's work when I first read it; I first encountered it as an undergrad. A.K. Dewdeny's writeup is reprinted in his "The Magic Machine: A Handbook of Computer Sorcery". Thanks for posting.
posted by grimjeer at 6:22 PM on January 18, 2010


I tend to think of Human Cognition in the terms and language supplied by Particle Physics;
Human cognition seems to me to be like the waveform of Cognition, in a comparison with the purpose-specific 'Point in Space' or, I guess, Particles of Cognition which has been the domain of a lot of robotics. Each "lineage" of robots in a series of experiments by specific labs being sort of walled off from it's competing projects (something that didn't happen with Humans as we Evolved, there weren't the jarring “jumps” that we see with robotics... in the Evolutionary transitional terms, from Pierolapithecus catalaunicus all the way to Mitochondrial Eve (the one from battlestar galactica, obviously ; ) and Modern Humans, what seems to be essential was needing to interact with a huge field of ideas, politically, and socially, without a common language, but a common set of experiences, the earliest Humans (like some of the newer ideas in robotics) benefited greatly from being able to absorb the ideas, learned techniques, and being able to incorporate the ideas from one group, or “project” (in Robotics terms), and to use those new ideas, or codes in ways that would change, and become “customized”, rather than starting from scratch.

I guess it is at this point when I recommend watching the Errol Morris created documentary Fast, Cheap, and Out of Control: (Transcript, of the movie which must be seen in film form for full effect). As I watched it, I felt like the film was able to break down the concept “Life”...like a part of the whole concept “Life”, the metonymy of Life expressed in Robotics, and the ... and the Synecdoche of Life for example by the taking the Life of a man who devoted his life to creating animal life from Shrubbery (Topiaries?) … I am not describing clearly my feelings of epiphany on those words that I felt as I watched the film, but seem unable to describe well right now at all, but it was clear as day to me as each person spoke of their passions, and it became clear how many forms that can arise out of one human concept, or in this case, word, how our experiences shape us, and how small events can hold no meaning to outsiders, but are the everything and all to that one individual; but I highly recommend it, especially while considering the definitions of those two tricky and under-rated concepts.
Sorry for the off topic, semi-rando-pinions, I very much enjoyed the post!
posted by infinite intimation at 9:08 PM on January 19, 2010


« Older Wu Tang Vs. The Beatles. You know you want it....  |  The Book Pirates of Peru. A sl... Newer »


This thread has been archived and is closed to new comments