Curious Frenchman who loves to make proof of concepts to learn and explore.
Published
Edited
Jul 6, 2020
2 forks
1 star
titleMd = md`# The Computational Boundary of a “Self”, an explorable explanation
Article https://www.frontiersin.org/articles/10.3389/fpsyg.2019.02688/full`
md`## Parameters`
md`### Defining the size of a square environment`
SpotsPerEnvironmentalGridSide = 10
md`### Generating an environment without any resource`
EnviromentalStateBase = new Array(SpotsPerEnvironmentalGridSide*SpotsPerEnvironmentalGridSide).fill(0);
md`### Probably for a spot in the environment to have exploitable resource`
viewof ratioOfFilledResources = html`<input type="range" min="0" max="1" value="0.3" step="0.05">`
ratioOfFilledResources
md`### Probably for spot in the environment to have an element`
viewof ratioOfFilledElements = html`<input type="range" min="0" max="1" value="0.5" step="0.05">`
ratioOfFilledElements
md`### Maximum amount of resource an element can exploit (per tick)`
viewof maxConsumption = html`<input type="range" min="0" max="1" value="0.02" step="0.01">`
ratioConsumptionFromEnvironment = maxConsumption / 0.5
radiusOfCommunication = 1;
md`## Defining environment and elements`
md`### Adding exploitable resources in the environment`
// Initialise State
EnviromentalState = EnviromentalStateBase.map( i => Math.random() < ratioOfFilledResources ? Math.random()*1 : 0)
md`### Adding elements in the environment`
ElementsState = EnviromentalStateBase.map( i => Math.random() < ratioOfFilledElements ? Math.random() : 0)
maxEnergyStored = 1
thresholdToGive = 0.7
viewof energyToGiveAtCommunication = html`<input type="range" min="0" max="1" value="0.01" step="0.001">`
energyToLostAtEachStep = 0.01
md`### Behavior of the element in its environment (called once per tick)`
updateElement = function(e,i) {
// life spending
e = Math.max(0, e - energyToLostAtEachStep);
if (!e) { ElementsState[i] = e; return }
// Consuming resources (non depleating environement)
if(EnviromentalState[i]*255 > 1 && e+ratioConsumptionFromEnvironment<maxEnergyStored) {
e += Math.min( maxConsumption, EnviromentalState[i]*ratioConsumptionFromEnvironment );
}
// messages are energy
let eltJ = Math.floor(i / SpotsPerEnvironmentalGridSide);
let eltI = i % SpotsPerEnvironmentalGridSide;
for(let ix = -radiusOfCommunication; ix <= radiusOfCommunication; ++ix) {
for(let jy = -radiusOfCommunication; jy <= radiusOfCommunication; ++jy) {
if(!ix && !jy) break;
if(ElementsState[((eltJ+jy)*SpotsPerEnvironmentalGridSide) + (eltI+ix)] > thresholdToGive) {
// ElementsState[((eltJ+jy)*SpotsPerEnvironmentalGridSide) + (eltI+ix)] = 0
// used for test, works
}
}
}
// Communications could be modelised through mailbox
// in this case directly giving or taking messages (here energy) is similar and quicker to implement
ElementsState[i] = e;
}
md`Trying to modify the canvas to be a side overlay for faster testing.
(but failing because ... iframe?!)`
html`<style>#defaultCanvas0DISABLED {position:absolute;top:-100px;}</style>`
md`## Display of the environment with elements according to the defined parameters`
canvas = p5(s => {
s.setup = () => {
s.createCanvas(800,400);
}
s.draw = () => {
var spotSize = s.height/SpotsPerEnvironmentalGridSide
// A]UPDATE
// 1) Update the elements
ElementsState.map( (e,i) => {
updateElement(e,i);
})
// 2) check if mouse is hovering over an element
let eltHover = s.createVector(-100,-100);
if(ElementsState[Math.floor( s.mouseY / spotSize ) * SpotsPerEnvironmentalGridSide
+ Math.floor( s.mouseX / spotSize )] > 0) {
eltHover = s.createVector(0.5 *spotSize + spotSize * Math.floor( s.mouseX / spotSize ),
0.5 *spotSize + spotSize * Math.floor( s.mouseY / spotSize ));
}
// B] DRAW
s.background(240)
// 1) Line Grid
s.noFill()
s.stroke(0)
for(var i=0; i<SpotsPerEnvironmentalGridSide+1; ++i) {
s.line(0, i*s.height/SpotsPerEnvironmentalGridSide, s.height, i*s.height/SpotsPerEnvironmentalGridSide);
s.line(i*s.height/SpotsPerEnvironmentalGridSide, 0, i*s.height/SpotsPerEnvironmentalGridSide, s.height);
}
// 2) Draw environement
EnviromentalState.map( (e,i) => {
let x = i % SpotsPerEnvironmentalGridSide * spotSize
let y = Math.floor( i / SpotsPerEnvironmentalGridSide ) * spotSize
s.noStroke()
s.fill(e*255)
s.rect(x,y,spotSize,spotSize)
})
// 3) Draw elements
ElementsState.map( (e,i) => {
let x = i % SpotsPerEnvironmentalGridSide * spotSize
let y = Math.floor( i / SpotsPerEnvironmentalGridSide ) * spotSize
s.stroke(0,0,100)
s.fill(e*255, 0, 0)
if(e) {
s.ellipse(x + spotSize*0.5, y + spotSize*0.5, spotSize*0.75, spotSize*0.75)
}
})
// 4) Draw communication radius of element
s.noStroke()
s.fill(0, 0, 200, 80);
s.ellipse(eltHover.x, eltHover.y, spotSize*1.1,spotSize*1.1)
}
})
visualsMd = md`# Visuals from the paper`
md``
md``
md``
md`## To do
1. ~~setup p5js~~
1. ~~generate visual grid~~
1. ~~black lines~~
1. ~~colored cells~~
1. ~~noisy grid based on 1 color~~
1. define 1 cell
1. must clarify the homeostatic aspect
1. e.g. could have a maximum consumption
1. display the cognitive light cone diagram, in time and space, of each element
1. interaction to add cell(s)
1. update/delete cell(s)
1. define behavior of a cell
1. related to the environment (grid)
1. related to other cells
1. communication probability
1. map visual differentiation per grid e.g.
1. grid 1 = color as analogy for cells
1. grid 2 = shape as analogy for bactery
1. etc clarify the link between grid N-1 / grid N / grid N+1
1. clarify abstraction across grids shwocasing the importance of
1. homeostatis
1. cognition
1. collaboration
1. consider how it can be integrated
1. e.g. viewof with slider
1. cf past work https://glitch.com/edit/#!/experiences?path=README.md testable https://experiences.glitch.me/configurator.html explained in video https://peertube.video/videos/watch/181e20bd-e9b1-4a05-867d-d54b84734191
`
md`# Imports`
import {p5} from '@makio135/p5'
md`# Lessons learned pair prog
## Session June 26
- insure that not too much time is spend on tool discovery
- take time for external communication e.g. Twitter
- refactoring code for next session
- refactor cells in Observable
## Session July 3
- nearly no time wasted on the tools
- better balance on goal vs implementation
- could write down goals
- have a separate prior 30min session on decided what will be communicated
## Chat July 6
- quotes from the paper
- behavior
- video animation
- call for action prompting an exploration with a specific concept to learn about
- e.g. what is homeostatis
`
md`# Related references
- [The Case Against Reality, Why Evolution Hid the Truth from Our Eyes by Donald Hoffman](https://wwnorton.com/books/9780393254693)
- could be interesting on the perception aspect of each element
- [Controllability of complex networks](https://www.nature.com/articles/nature10011)
- Metzinger on self-models
- kappa https://kappalanguage.org
`
Purpose-built for displays of data
Observable is your go-to platform for exploring data and creating expressive data visualizations. Use reactive JavaScript notebooks for prototyping and a collaborative canvas for visual data exploration and dashboard creation.
