Version 1.1

Added load and save for model parameters and outputs.
Updated documentation.

Author: Caspar Addyman

GAMIT Matlab Code.docx

@@ -1,4 +1,4 @@
-function ReferenceCurve = GAMIT_Lifetime(params, N, showGraphics)
+function ReferenceCurve = GAMIT_Lifetime(params, N, showGraphics, exportRawData)
 %
 % generates an average lifetime activation curve. Has default values but
 % accepts overrides
@@ -12,6 +12,9 @@
 if nargin < 3
     showGraphics = false;
 end 
+if nargin < 4
+    exportRawData = true;
+end
 %initialise storage
 gamitScore =zeros(N,params.nIterations); 
 scoreDelta =zeros(N,params.nIterations); 
@@ -74,6 +77,14 @@
 ReferenceCurve.GamitScoreUncertainty = std(gamitScore, 1);
 ReferenceCurve.Delta = mean(scoreDelta,1); 
 
+if exportRawData
+    Score = ReferenceCurve.GamitScore';
+    ScoreUncertainty = ReferenceCurve.GamitScoreUncertainty';
+    Delta = ReferenceCurve.Delta';
+    t = table(Score,ScoreUncertainty,Delta);
+    writetable(t,'GAMIT_Lifetime.csv','Delimiter',',');
+    save('GAMIT_Lifetime_Params.mat','params');
+end
 
 
 if showGraphics

GAMIT_Lifetime.m

@@ -1,28 +1,36 @@
-function params = GAMIT_Params()
+function params = GAMIT_Params(matfile)
 %
 % helper function to return standard GAMIT spreading activation parameters
+% if matfile parameter is passed we load params from corresponding .mat
+% file
 
-%curve evolution parameters
-params.initialActivation = 1;
-params.nColumns = 200;
-params.nIterations = 2000;      % number of timesteps we evolve curve for
+if nargin == 1
+    temp = load(matfile);
+    params = temp.params;
+else    
+    %curve evolution parameters
+    params.initialActivation = 1;
+    params.nColumns = 200;
+    params.nIterations = 2000;      % number of timesteps we evolve curve for
 
-params.alpha = 0.7;             % self activation 
-params.beta = 0.14952;          % spreading activation
-params.noiseFactor = 0.000025;  % noise
+    params.alpha = 0.7;             % self activation 
+    params.beta = 0.14952;          % spreading activation
+    params.noiseFactor = 0.00025;  % noise
 
-params.GaussianFit = false;      %is gamit score the stddev of the best fit gaussian 
-                                 %or is it SummedActivation + MaxActivation?
+    params.GaussianFit = 0;      %is gamit score the stddev of the best fit gaussian 
+                                     %or is it SummedActivation + MaxActivation?
 
-%curve sampling params
-params.bias = 0.87;              %single parameter to account for fact that humans always underestimate/overproduce intervals.
-params.sampleErrorSize = .02;    % margin of error on an sample from curve
-params.MemoryUncertainty = false; %is there any additional uncertainty when we read off from the lifetime Curve?
+    %curve sampling params
+    params.bias = 0.87;              %single parameter to account for fact that humans always underestimate/overproduce intervals.
+    params.sampleErrorSize = .05;    % margin of error on an sample from curve
+    params.MemoryUncertainty = 0; %is there any additional uncertainty when we read off from the lifetime Curve?
 
-%prospective model parameters
-params.WorkingMemoryDelta = false; %Do we just a subset of sampling deltas or all of them?
-params.RandomAccessMemory = false; %If WMD=true, is the subset random or just the most recent deltas
-params.nSampleDeltas = 6;       % how many samples can we keep in memory?
+    %prospective model parameters
+    params.WorkingMemoryDelta = 0; %Do we just a subset of sampling deltas or all of them?
+    params.RandomAccessMemory = 0; %If WMD=true, is the subset random or just the most recent deltas
+    params.nSampleDeltas = 6;       % how many samples can we keep in memory?
 
-params.sampleFrequency = 50;    % base rate of one sample every fifty ticks
-params.PoissonSampling = true;  % are time points samples according to a Poisson process or uniform random variable
+    params.sampleFrequency = 50;    % base rate of one sample every fifty ticks
+    params.PoissonSampling = 1;  % are time points samples according to a Poisson process or uniform random variable
+
+end

GAMIT_Params.m

@@ -1,4 +1,4 @@
-function GAMIT_Retro_Pro_Interaction(targetTime,nSamples,lowCognitiveLoad,highCognitiveLoad,showGraphics)
+function GAMIT_Retro_Pro_Interaction(targetTime,nSamples,lowCognitiveLoad,highCognitiveLoad,showGraphics,exportRawData)
 %
 % Demonstrate how the retrospective and prospective time estimates are
 % calculated in the GAMIT model.
@@ -7,22 +7,24 @@ function GAMIT_Retro_Pro_Interaction(targetTime,nSamples,lowCognitiveLoad,highCo
     targetTime = 600;
 end
 if nargin<2
-    nSamples = 50;
+    nSamples = 20;
 end
 if nargin<3
     lowCognitiveLoad = 0.95;
-    highCognitiveLoad = 1.1;
+    highCognitiveLoad = 1.05;
 end 
-if nargin<4 
+if nargin<5 
     showGraphics = true;
 end
-relativeEstimates = true;
+if nargin<6
+    exportRawData = true;
+end
+relativeEstimates = true; % divide through by targetTime
 
 %step 1: Get default params & generate a reference curve
 params = GAMIT_Params();
 lifetimeCurve = GAMIT_Lifetime(params);
 
-
 testTimes = targetTime * ones(1,nSamples);
 
 %%% RETROSPECTIVE %%%
@@ -33,6 +35,11 @@ function GAMIT_Retro_Pro_Interaction(targetTime,nSamples,lowCognitiveLoad,highCo
 prospectiveLow = GAMIT(testTimes,lowCognitiveLoad,true,false,params,lifetimeCurve);
 prospectiveHigh = GAMIT(testTimes,highCognitiveLoad,true,false,params,lifetimeCurve);
 
+if exportRawData
+    t = table(retrospectiveLow,retrospectiveHigh,prospectiveLow,prospectiveHigh);
+    writetable(t,'GAMIT_Retro_Pro_Interaction.csv','Delimiter',',');
+    save('GAMIT_Retro_Pro_Params.mat','params');
+end
 
 if showGraphics
     %Graphics code shows interaction plot like Block, Hancock & Zakay 2010
@@ -43,10 +50,11 @@ function GAMIT_Retro_Pro_Interaction(targetTime,nSamples,lowCognitiveLoad,highCo
     %get retrospective 'coordinates'
     yR = [mean(retrospectiveLow),mean(retrospectiveHigh)];
     yeR = [std(retrospectiveLow),std(retrospectiveHigh)];
+    yeR = yeR /sqrt(length(retrospectiveLow));
     %get retrospective 'coordinates'
     yP = [mean(prospectiveLow),mean(prospectiveHigh)];
     yeP = [std(prospectiveLow),std(prospectiveHigh)];
-    
+    yeP = yeP /sqrt(length(prospectiveLow));
     if relativeEstimates
         yR = yR/targetTime;
         yeR = yeR/targetTime;
@@ -56,14 +64,15 @@ function GAMIT_Retro_Pro_Interaction(targetTime,nSamples,lowCognitiveLoad,highCo
 
     hold on;
     %plot lines
-    line(xC, yR,'Color','b');
-    %offset this xcoord slightly for ease of view
+    %offset second set of xcoords slightly for ease of view
     xCprime = xC+diff(xC)*.02;
-    line(xCprime, yP,  'Color','r');
-    legend('Retrospective','Prospective');
+%     line(xC, yR,'Color','k','LineStyle','--','Marker','o');
+%     line(xCprime, yP,  'Color','k','Marker','o');
     %plot error bars
-    errorbar(xC,yR,yeR,'Color','b');
-    errorbar(xCprime,yP,yeP,'Color','r');
+    errorbar(xC,yR,yeR,'Color','k','LineStyle','--','Marker','o');
+    errorbar(xCprime,yP,yeP,'Color','k','Marker','o');
+    legend('Retrospective','Prospective');
+ 
     hold off;
     xlabel('Relative Cognitve Load (Normal load = 1.0)');
     if relativeEstimates

GAMIT_Retro_Pro_Interaction.m

@@ -1,4 +1,4 @@
-function GAMIT_Weber_Demo(targetTime, nSamples, prospectiveFlag,method)
+function GAMIT_Weber_Demo(targetTime, nSamples, prospectiveFlag,method,exportRawData)
 %
 % Demonstrate that time estimates in the GAMIT model have the scalar
 % property. 
@@ -8,6 +8,7 @@ function GAMIT_Weber_Demo(targetTime, nSamples, prospectiveFlag,method)
 %   prospectiveFlag = 0,false [Retrospective], 1,true [Prospective]
 %   method          = 0,'hist' compare targetTime and 1.5*targetTime as histograms
 %                     1,'bars' compare targetTime & 0.2:0.2:2 * targetTime as estimates & errorbars
+%   exportRawData   = 0,1,true,false, save 
 
 if nargin<1
     targetTime = 800;
@@ -21,6 +22,9 @@ function GAMIT_Weber_Demo(targetTime, nSamples, prospectiveFlag,method)
 if nargin<4 
     method = 0;
 end
+if nargin<5
+    exportRawData = false;
+end
 
 if prospectiveFlag 
     pro_or_ret = 'Prospective';
@@ -65,6 +69,13 @@ function GAMIT_Weber_Demo(targetTime, nSamples, prospectiveFlag,method)
     subplot(1,2,2);
     bar(x,[n' n2']);
     title(['Scaled histograms of estimates - ' pro_or_ret]);
+    
+    if exportRawData
+        t = table(retrospectiveLow,retrospectiveHigh,prospectiveLow,prospectiveHigh);
+        writetable(t,'GAMIT_Weber_Demo.csv','Delimiter',',');
+        save('GAMIT_Weber_Params.mat','params');
+    end
+        
 else %webers law as error bars. 
     multipliers = [0.2:0.2:2];
     nSteps = length(multipliers);