function [pE,pC] = spm_ssr_priors(pE,pC)
% augments prior moments of a neural mass model for CSD analyses
% FORMAT [pE,pC] = spm_ssr_priors(pE,pC)
%
% pE - prior expectation
%
% adds
%
% input and noise parameters
%--------------------------------------------------------------------------
%    pE.a - neuronal innovations         - amplitude and exponent
%    pE.b - channel noise (non-specific) - amplitude and exponent
%    pE.c - channel noise (specific)     - amplitude and exponent
%    pE.d - neuronal innovations         - basis set coefficients
%
%--------------------------------------------------------------------------
%
% pC - prior (co)variances
%
% Because priors are specified under log normal assumptions, most
% parameters are simply scaling coefficients with a prior expectation
% and variance of one.  After log transform this renders pE = 0 and
% pC = 1;  The prior expectations of what they scale are specified in
% spm_lfp_fx
%__________________________________________________________________________
%
% David O, Friston KJ (2003) A neural mass model for MEG/EEG: coupling and
% neuronal dynamics. NeuroImage 20: 1743-1755
%__________________________________________________________________________
% Copyright (C) 2008 Wellcome Trust Centre for Neuroimaging
 
% Karl Friston
% $Id: spm_ssr_priors.m 5816 2013-12-23 18:52:56Z karl $
 
% catch
%--------------------------------------------------------------------------
try, pE.L; catch, pE.L = 1; end
 
% number of LFP channels and sources (endogenous inputs)
%--------------------------------------------------------------------------
if size(pE.L,1) == 1, n = size(pE.L,2); else, n = 1; end
if size(pE.C,1),      m = size(pE.C,1); else, m = 1; end
 
% add prior on spectral density of fluctuations (amplitude and exponent)
%--------------------------------------------------------------------------
pE.a = sparse(2,m); pC.a = sparse(2,m) + 1/128; % neuronal fluctuations
pE.b = sparse(2,1); pC.b = sparse(2,1) + 1/128; % channel noise non-specific
pE.c = sparse(2,n); pC.c = sparse(2,n) + 1/128; % channel noise specific
 
% neuronal innovations (DCT coefficients for structured spectra)
%--------------------------------------------------------------------------
d    = 8;
pE.d = sparse(d,m); pC.d = sparse(d,m) + 1/128;