What counts as cheating in Monty?

General
1

Ive been dicking around with monty for a couple weeks now. As a tech/engineer I find myself doing all kinds of fun stuff at the interface between sense and cortex, world and organism-in-world.

If Monty is just a cortex with a format for inputs, what am I allowed to do on the ins and outs? Obviously the brain is more than the cortex, and from what I can see any one sensory organ is far more complicated than the cortex offered from Monty. And that is as it should be: TBT says at that level the alg is the same, over and over, shared between columns. This is the easy part when you compare it to a cochlea or an eye. It’s a lot harder to get the motor response to click with a sense than it is for a sense to click with a brain.

So, between the stock eyeballs and ears I find, and the stock cortex (Monty), what am I allowed to do between these two interfaces that would not be… cheating? I have already decided that its halting is a non-starter. Is the aim to do what animals do, or is the aim to make Monty make perfect instantaneous decisions? If there is an evolutionary principle to have as a rule (dont die, for instance, one that is not in Monty currently), am I allowed to code it in and see what falls out? How stable is the monty repo, and is it going to change on me in the next month in a way that will make all the fun stuff I’m doing now totally irrelevant?

Thats my only question, stated in a whole bunch of questions. Thanks a thousand brains over for working on this. Its been nothing but fun for me.

1 Like
2

For the repo question, you got here just in time​:joy: see https://forum.thousandbrains.org/t/2026-07-upcoming-and-ongoing-engineering-platform-changes/1188 for the upcoming changes

1 Like
3

Here’s something on the other question The road to a Generally-Intelligent Monty? - #9 by vclay

1 Like
4

This is a link from a reply to a post here (I’ll track the post down again) and the keynote speaker is the person who rewired a ferret brain, swapping the eyes and ears, and the brain reorganized itself.

1 Like
5

Found the post

1 Like
6

o man these are great thank you for the links!!!

7

The repo makes breaking changes pretty much every week, but we do list what we break. Sometimes these are small changes like removing or changing a method somewhere. Sometimes, the main parts of Monty are changed.

We follow semantic versioning, and the MAJOR version 0 is intended to communicate that we intend to break things.

We will use MAJOR version number 0 for the ongoing development of Monty. Incrementing the MAJOR version number beyond 0 should be a future RFC, as it will be a significant event.

– RFC 7 - Monty Versioning

Things will not stabilize until MAJOR version 1.

We do publish numbered versions frequently, though, so that if you pin to a specific Monty version, e.g., v0.50.0, then your code should continue to work. For example, our repository accompanying the “Thousand Brains Systems: Sensorimotor Intelligence for Rapid, Robust Learning and Inference” paper, is pinned to tbp.monty==0.1.0 :slight_smile:. Many code changes since then.

8

excellent, thank you so much!

9

Here’s the ferret paper from 1992 for those interested: https://www.jneurosci.org/content/jneuro/12/9/3651.full.pdf

1 Like