COMMENTS ON THE PVMODEL VERSION 0.9 SCRIPTS
	Last revised 10/8/97, rms


This is the not-quite-ready-for-prime-time implementation of the
position-variable model, written in MATLAB.  The code was rewritten
in MATLAB because the NeXT OS has virtually disappeared from the 
horizon, and because the NeXT implementation had become too complicated.


The present scripts were written to describe the stimuli in the
Trahiotis and Stern (1994) JASA letter to the editor, and these scripts
have been used to generate the figures for the ISH-97 paper (Stern and 
Trahiotis, 1997).  

The present scripts can be used for the following stimuli:

	* Pure tones
	* AM tones
	* Triple AM tones
	* Any combination of tones, such as the stimuli of Hill and Darwin
	  (simultaneous onsets only)
	  
To run the scripts, type "pvmodel" in the MATLAB Command Window.  

The specific parameters of the implementation are presently determined by
variables in subroutines.  I have tried to modularize the scripts to the
extent possible, so that new stimuli or experiments can be run without
disrupting the main code (which is rather unintelligible in spots), but I
have not been totally successful in this regard yet.  

Here is a summary of the subroutines called and their functions:

circconv.m		implements circular convolution (needed in 
			    the straightness weighting)
display.m		used to generate the meshplots
freqgen.m		generates the vector of CFs used for calculations				
f_init.m		contains parameters pertaining to frequencies of 
			    calculation
hm.m			generates frequency response of auditory filter 
			    (according 
			    to Siebert's early model, as used by Colburn (1973)
ptau.m			generates p(tau,fc) function 				
stim_init.m		contains values of stimulus parameters
str_init.m		contains values of straightness parameters
sttcomp.m		script that does all the stimulus-specific calculations
t_init.m		contains temporal parameters pertaining to calculation


Here is a summary of the stimulus parameters and their values used in this 
implementation.  Values below are used to generate the upper panel of
Figure 2 in Stern and Trahiotis (1997).  

[In stim_init:]

stimset = 1;   	% stimset must be 1 or 2 (for the triple-AM stimuli, as defined
		% in Stern and Trahiotis (1994, 1997))
As = [1 1 1]; 	% component amplitude
f0 = [.3 .5 .7];% component frequency in kHz
ts = 1.5; 		% waveform time delay, ms
fm = .025; 		% modulation frequency, kHz
modindex = 1; 	% modulation index



And here is a summary of the infrastructure parameters and their values used in 
this implementation:

[in f_init:]

fl = .2;		% lower frequency for calculations (in kHz)
fcent = .5;		% center frequency, which calculations are forced to include
fh = .9;		% upper frequency for calculations		
finc = .05;		% frequency ratio (determines spacing of frequencies)


[in t_init:]

tinc = .05; 		% increment for t and tau in msec; originally .05
tlimit = 40;		% max time in msec; must be a multiple of the period
taulimit = 3; 		% limit for calculation in msec


[in str_init.]

strnum = 5;   	% number of CFs; must be odd and > 1
strflag = 0; 	% 0 for no str wtg; 1 for str wtg


Additional comments:

To display mesh plots for the responses of the coincidence counters,
execute the statements in display2d.m.  (I currently copy and paste these 
into the main workspace.)  There is a line that can be uncommented if
plots with weighting by p(tau,fc) are desired.  Stern and Trahiotis (1997)
makes use of a different plotting routine, which produces plots in 
slightly different form.

Non-normalized predicted lateral position is computed at the end of the 
pvmodel script.


Known bugs:

	* Currently the delta parameter described in Stern and Trahiotis (1997)
	  must be set equal to finc.