<filter>
The filter object defines a linear filter.
2-D and 3-D filters
Gabor
This Gabor filter is a 3-D spatiotemporal filter used in the motion energy (me_gabor_01) and binocular disparity energy (binoc_filter) models.<filter> type Gabor # type of filter dimension 3d direction 0 (deg) # preferred direction sf 1.25 (cyc/deg) # Spatial frequency ssd 0.18 (deg) # SD space tf 0.0 (cyc/s) # Temporal frequency tsd 0.015 (s) # SD time tilt_type none # ["none"] "rotate" "shift" tilt_frac 0.0 # Amount of rotation, 0-none, 1-vel. write_filter 0 # 0-none, 1,2,3,4 or 5=all opponent no </filter>Gabor_CWQ
This filter is used in the BDE_CWQ family of models, based on the work of Chen et al. 2001, that can be found here.<filter> type Gabor_CWQ # Type of filter dimension 3d direction 0 (deg) # preferred direction sf 1.2 (cyc/deg) # Spatial frequency sd_orth 0.267 (deg) # SD space, orthogonal to sinusoid sd_par 0.4 (deg) # SD space, parallel to sinusoid phase 0.0 (deg) # Spatial phase offset phase_shift 180.0 (deg) # Phase shift of Right Eye alpha 2 # Temporal filter skewness: 2,3,... tau 0.060 (s) # Temporal filter time constant tf 2.0 (cyc/s) # Temporal filter cosine TF phi 18.0 (deg) # Temporal filter cosine phase eta 1.0 # Fraction of Quadrature filter to add write_filter 0 # 0-none, 1,2,3,4 or 5=all </filter>reich
This filter is used in Reichardt detector (reichardt-type) models.<filter> name reich dimension 3d lp_tau 0.010 # Low-pass temporal filter time constant (s) hp_tau 0.500 # High-pass temporal filter time constant (s) ### Set the number of Reichardt Detector units for spatial integration n_unit_x 20 # Number of units along x-dimension n_unit_y 1 # Number of units along y-dimension dx 1 # x-offset between paired receptors (sampling units) opp_flag 1 # 1-opponent RD is subtracted, 0-non-opponent dump 0 # 1-write traces for center RD unit to 'zzz.rd.pl' </filter>
1-D filters
These are a set of 1-D temporal filter used in lin_01 models, and also to define temporal filters for some 3-D models.
The examples below show all possible parameters that can be defined, but some parameters are mutually exclusive in their actions. For example, either max or area should be used, but not both. Some parameters are optional, such as n, which is typically set from the model tn value.
boxcar
Boxcar function with optional 2nd lobe, which follows immediately after the first.<filter> type boxcar # Shape of filter dimension 1d # Dimension of filter n 100 # Length of filter (sampling units) t0 20 # start time (sampling units) tn 10 # width of boxcar (sampling units) amp 0.002 # amplitude tn2 20 # width of optional 2nd lobe (sampling units) amp2 -0.001 # amplitude of optional 2nd lobe plot_filename NULL # Name of file for writing plot data [NULL] </filter>exp
One-sided decaying exponential function, with optional sum of 2nd exponential, and optional smoothing w/ a Gaussian<filter> type exp # shape of filter dimension 1d # Dimension of filter n 100 # length of filter (sampling units) tau1 0.004 (s) # time constant of decaying exponential tau2 0.008 (s) # optional 2nd exponential, time constant. amp2 -1.000 # relative amplitude of optional 2nd exponential max 0.5 # maximum value of filter [or specify 'area'] area 1.0 # total area of filter [or specify 'max'] plot_filename NULL # name of file for writing plot data [NULL] gauss_sd 0.0 # optional SD for Gaussian smoothing [0.0 - none] </filter>alpha
Alpha-function, with optional sum of 2nd alpha-function, and optional smoothing w/ a Gaussian.<filter> type alpha # shape of filter dimension 1d # Dimension of filter n 100 # length of filter (sampling units) tau1 0.004 (s) # peak time of first alpha function tau2 0.000 (s) # peak time of optional 2nd alpha function amp2 0.000 # relative amplitude of optional 2nd function [0.0 none] max 0.5 # maximum value of filter [or specify 'area'] area 1.0 # total area of filter [or specify 'max'] plot_filename NULL # name of file for writing plot data [NULL] gauss_sd 0.0 # optional SD for Gaussian smoothing [0.0 - none] </filter>Gaussian
Gaussian with optional sum of 2nd Gaussian.<filter> type Gaussian # shape of filter dimension 1d # Dimension of filter n 100 # length of filter (sampling units) mean 0.020 (s) # mean of Gaussian sd 0.002 (s) # SD of Gaussian mean 0.020 (s) # mean of optional 2nd Gaussian sd 0.005 (s) # SD of optional 2nd Gaussian amp2 -1.000 # relative amplitude of optional 2nd Gaussian max 0.5 # maximum value of filter [or specify 'area'] area 1.0 # total area of filter [or specify 'max'] plot_filename NULL # name of file for writing plot data [NULL] </filter>Maxwell
Maxwell function with optional 2nd lobe.<filter> type Maxwell # shape of filter dimension 1d # Dimension of filter n 100 # length of filter (sampling units) origin zero # ['zero'] or 'center' m1 0.020 (s) # peak time of Maxwell kernel m2 0.040 (s) # optional 2nd kernel, peak time amp2 -0.500 # relative amplitude of optional 2nd kernel max 0.5 # maximum value of filter [or specify 'area'] area 1.0 # total area of filter [or specify 'max'] plot_filename NULL # name of file for writing plot data [NULL] </filter>
WM Working Models www.iModel.org/wm