"mike" <mike-anderson at nwu.edu> wrote in message
news:mike-anderson-2110001610180001 at dhcp083174.res-hall.nwu.edu...
> Me again... This is a question in regards to a class that i am taking
> now, i would ask a TA but it is the weekend and they are unavailable.
>> The question is basiclly to devise a neural code to transmit a four letter
> word "axon" in 400ms (assuming no more than 15 action potentials every 100
> ms, so a total of 60 potentials is allowed, the refractory period of the
> potential is 5ms).
>> The way i understand coding is that multiple potentials, as well as the
> timing between them, encode information. I am also aware that there are
> different types of coding that have been hypothesized (temporal, place...)
> but the time/rate code is the most widely accepted. So basically, do i
> need to devise a code that will have a different number of potentials and
> times between them for each letter of the alphabet (i.e., A=1 potential,
> Z=26, with only the refractory period between potentials, and an extra
> time delay between letters?), or is there an easier way that i could do
> this?
>> I think this is a fairly basic problem, however, our text does not cover
> the encoding of information very well, and i can not seem to find helpful
> information on the net.
>> So if you have any pointers here please post them, also, if there are any
> references that i should check out i would be grateful for that info too.
>> thanks
> -mike
> In my dreams i'm dying all the time
>http://www.mp3.com/mikeanderson
I hope its a programming course because real neurons don't use neural
codes this way.
However, it only takes 7 bits of information to transmit one computer
character in the ASCII code (with 128 possible characters). Four
characters would require only 28 bits, easy enough with a system that
could potentially send up to 60.
If you restrict the character set to only the lower case English alphabet
(26 characters) would require only 20 bits to transmit the four
letter word.