[Neuroscience] Re: coding principles of the retina

codyfrew via neur-sci%40net.bio.net (by codyfrew from gmail.com)
Mon Mar 5 04:03:02 EST 2007


On Mar 4, 8:18 am, "J-P" <Cervell... from gmail.com> wrote:
> Hello great Brain Community
>
> Today i'd like to know more about the underlying coding Principles of
> the retina. I have read several papers and understood the basic
> principles but  where im stuck is "Dynamic predictive coding by the
> retina"* How can the receptive field of the ganglion cell dynamically
> adapt to  maximise response to novel stimuli. On slightly different
> topic i'd also like to know, where the melanopsin containing ganglion
> cells project to and what they do.
>
> thanks
> J-P
>
> *Dynamic predictive Coding by the retina; Toshihiko Hosoya  2005
> nature vol 436
>
> P.S.: What are biphasic spatio temporal receptive fields?

J-P, it's been a couple of years since my coursework- so I can't help
enough to answer all of your questions. I can however, answer your
first one. Predictive coding simply means that the cells are receiving
feedback from neurons higher up the stream in order to expect certain
patterns, thereby forcing them to be slightly more receptive in
certain reas and to certain spatio-temporal patterns be altering their
chemical structure slightly. ou can read more about this by googling
top-down processing, neural feedback, or Gestalt effects. Some Gestalt
effects are seen higher up in the neural 'stream' (perception), but
some are closer to sensation.

By the way, there is a book entitled <a href="http://www.amazon.com/
Intelligence-Jeff-Hawkins/dp/0805078533/
ref=pd_bbs_1/103-3892093-8283814?
ie=UTF8&s=books&qid=1173085223&sr=8-1">"On Intelligence" by Jeff
Hawkins</a> that can help you understand predictive neural coding much
better. I highly recommend it to everyone.

Dynamic just means that it is happening on-the-fly. It is not the same
feedback coming back to the cells every time (it is not static).
Everything has to be dynamic to adapt to changes in the environment,
which cannot be account for by instructions. This is one of many
problems encountered by Artificial Intelligence experts.

If I had to take a shot at your post-script question, I would break it
down and take my best shot. Biphasic is a term that refers to two
distinct patterns or frequencies. I would guess that biphasic spatio
temporal receptive fields are referring to a receptive field (a source
of input- for instance some cells in V1, the primary visual processing
area of our brain, only receive information from 1 degree of our
visual field. Therefore, that cell has a receptive field of 1 degree)
that is sensitive to two different frequencies of information that
happen to be encoding both spatial and temporal information. I would
guess that a cell with such a receptive field would have double-bell
shaped graph with two peaks, where the activation sensitivity is on
the y axis and frequency of spatio temporal patterns are on the
bottom.

Sorry if I have misconstrued that or have written in a confusing
manner. Any neuroscience-fresh readers, please correct any errors!
   Best,
      -Cody Frew



More information about the Neur-sci mailing list