Hi,

On Thu, Feb 13, 2014 at 1:54 AM, mariolakakis . <[email protected]>wrote:

> Consider A --> B --> C and A --> B --> D
>
> According to how CLA forms connections, A is followed by  B and B is
> followed by C and D. So, when B occurs the system will predict that C or D
> will occur next.
>
You're right up to this part.


> Since, a connection goes both ways the system will also predict A
>
I don't understand what you mean here..basically it is not true (as said
now).
Oh, maybe I got what you mean (?): The "arrows" A-->B etc represent
consequtive states in (temporal) sequence. Not "neuron axons" in brain/CLA.
Therefore the assumption "connections go both ways, implying also B-->A" is
false. Note that the knowledge (eg points "A", "B",..) are represented in a
distributed manner (see SDRs) - that is by eg 200 neurons.

Is that what you were thinking, or I'm completely wrong?

> but B --> A is a false prediction. How do you make that distinction? You
> could use arrows but I'm sure that's not how the brain works.
>
> Cheers, Mark


-- 
Marek Otahal :o)
_______________________________________________
nupic mailing list
[email protected]
http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org

Reply via email to