Parr, Pezzulo, Friston 2022 Textbook Cohort 6, Meeting 17, applying ActInf 4

Textbook Group meeting for Parr, Pezzulo, Friston 2022 .

Aug 19, 2024

▶ Watch on YouTube ↗

Session details

Date: Aug 19, 2024

Series: Parr, Pezzulo, Friston 2022 Textbook Cohort 6, Meeting 17, applying ActInf 4

Transcript

AI-generated transcript excerpt

The full transcript is available on GitHub. This excerpt is generated by automated speech recognition and may contain errors.

okay we're in cohort sixo question mark And discussing some applying active inference questions and Jeff has turned us to the inspiration topic so where were we Jeff or continue yeah so um several of us Dean tickles and Anna magdalina herado and um um Hector Manriquez we've been going back and forth for for about two weeks on on what inspiration means uh through the lens of active inference and and for me inspiration and I think this is a really good um definition and I think Dean offered this up and it was um what happens when you have a new new model or a new artifact that has no through line no uh contiguity for how you receive how you arrived at that specific phenomena or or or experience or or bit of information right it is it is that that is seemingly without cause Okay lot of ways we can model it just are there any more General comments or or definitions or or thoughts and then otherwise we're g to kind of walk this pseudo chapter 6 path for H how we would move towards structuring it you're basically asking like how does like Insight arise into like the shape of some like object you're looking at or some insight you're looking at is that what that last question Zach yeah this uh this reminds me of really core active entrance right whenever you're talking about like if you just have a whole bunch of particles you can differentiate the states of the particles are in by the states that seem to have this kind of sub thing going on and the maintenance of the sub thing and this thing want by virtue of existing very core definition of active inference right or I'm sorry the uh I'm forgetting the word of it though it's a very important word the idea that because it exists we can talk about the Dynamics of it existing I'm sorry I can rephrase that if you I did not that very conditioning conditioning modeling of something on that thing being like existing or measured yes absolutely and so it's going to maintain its own state but the question I have is could you guys describe the no three line or continuity because by saying that it is a thing aren't we already talking about it has a thingness about it we are already talking about some states where it is instead of some states where it is not yeah like even a really in inspired new line of poetry it's like it it may have a chain of causation so then if we were if we were um tasked to make an active inference model um and in that first question chapter six like what system are we modeling it's like what what question are we trying to explore are we trying to reify our definition or observation that inspiration is that that arrives at with without uh uh cause or are we we looking to quantify like some index of inspiration where on the lower end there's like complete continuity and then on the upper end there's less continuity here's one Boulder paper with the kind of more on the the Insight [Music] um okay so from chapter six we have those four stages of the question let's just answer the first question and then we'll look over to to a few other um kind of uh tools for modeling but what what system are we going to model like can we pick some specific inspiration setting just to make it clearer a song comes to an artist in a dream okay here I'll I'll I'm going to pop over to this other um active block frence Koda just because um it it complements the chapter six so in the textbook chapter six gives these four steps um that suffice to design active inference generative model so big picture the first part is like system of Interest identification systems Engineering Systems modeling then you get into the kind of chapter 4 chapter seven and eight questions like is it going to be a discrete time or continuous time what kinds of State spaces are the variables and then question three and four are about setting up basically the agents and the niche and it's this is dealt with at a pretty overview level in the textbook and um depending on how how complicated the setting is that this can be…