Angelette Adjunctive Health Services
92 Market Street
Balranald, NSW 2715
to see how initial states impact reasoning follow directions at botton of page.
Note: ARU models were originally optimized for an earlier version of Google Chrome or internet explorer 7. with Java enabled. To use them in a contemporary version of the Chrome browser requires installing of the IE Tab15.1.24.1 extension available through Chrome as well as some tweeking of Java itself. One may find it easier to download NLOGO itself and open Models locally on your Computer.We are working to convert these models to HTML5. Thank you for your patience.
No two Minds are born exactly alike. Even genetally duplicate twins, although exactly alike in many other respects, are individuals. And some differences in judgment can be expected. one twin child may say 3 is greater than 2 yet the other twin may get it wrong. Computers on the other hand seem alway to get the same answer.
One might be inclined to object that it must be impossible to model my mind. It is unique!
The networks CAN GET THINGS WRONG for the same reasons that we do: initial neural state may be too varied from a norm.
But, we can demonstrate that networks arrive at similar conclusions within a broad range of different initial conditions.
It is true that old fashioned computers could never hope to capture this uniqueness feature. But these networks function entirely differently than computers.
click read file. click balance weights. click go.
click go again to stop.
click display output for detailed result.
Circles indicate high activation interpret as accept.
Triangles indicate low activation interpret as reject.
Default Activation: is initially a single ideal value that changes in response to node connectivity representing Strict Mind Brain Identity.
Randomize Activation: initially random values that change in response to node connectivity representing uniquely initial bran states.
click read file again. and balance weights. click randomize activation. and go.
Result, different initial randomized set of neurons each time you repeat this process.
Like individual minds, as many times as not networks will reach similar conclusions.
The closer to an ideal initial activation the more likely the rational outcome.
But rationality isn't guaranteed for our models anymore than it can be for our twins.
Conclusion: Mind Brain Identity
does not require Strict Identity.
The more neurons we add and connect the "smarter" the networks become.
[*] Copyright 2010 Project Selector Interface - Dir. Willow Angelette . All rights reserved. Android app NavigationPanel
Angelette Adjunctive Health Services
92 Market Street
Balranald, NSW 2715