In recent years the increasing availability of multi-electrode recordings has led to the application of neural decoding techniques to the recovery of complex stimuli such as natural scenes. A linear decoding algorithm was presented in  for the reconstruction of natural scenes with recognizable moving objects using recordings from a neural population of the cat’s Lateral Geniculate Nucleus (LGN).
Most of the current models of encoding in the early visual system (retina, LGN, V1) consist of a linear receptive field followed by a non-linear spike generation mechanism. In  we considered a neural circuit architecture consisting of receptive fields in cascade with an equal number of spiking neural circuits. The neural circuits investigated were integrate-and-fire neurons and ON-OFF neurons with random thresholds and feedback. We demonstrated for the first time a decoding algorithm for natural scenes and shown its dependence on the noise level.
We investigate a neural encoding architecture for visual stimuli consisting of classical receptive fields (center surround or Gabor) in cascade with an ensemble of Hodgkin-Huxley neurons. Recovery of stimuli encoded with an ensemble of Hodgkin-Huxley neurons with known phase response curves was achieved based on the I/O equivalence between Hodgkin-Huxley neurons and Project-Integrate-and-Fire neurons in . The ensemble of Hodgkin-Huxley neurons considered here is assumed to have unknown phase response curves . We provide a visual stimulus reconstruction algorithm based on the spike times generated by the ensemble of Hodgkin-Huxley neurons and demonstrate its performance using natural video sequences (movies). Fig. 1 shows a sample time instant (a frame of a movie) of the reconstructed (left) and the original (right) visual stimulus.
The work presented here was supported by AFOSR under grant number FA9550-09-1-0350.
IEEE Transactions on Information Theory 2010., 56(2)
to appearPubMed Abstract | PubMed Central Full Text
In In Phase Response Curves in Neuroscience, Springer Edited by Nathan W. Schultheiss, Astrid Prinz, and Rob Butera. 2010.