Matlab Class Chapter 4.pdf

Chapter 4 - 1 - 4/18/2007

Chapter 4 - Beginning the first experiment

In this chapter we are actually going to begin writing real code for a real experiment. We

are going to do a version of a very cool illusion first demonstrated by Mike Webster

(Webster, Kaping, Mizokami, and Duhamel, 2004, Adaptation to natural facial

categories. Nature 428, 558-561.)

The illusion works like this – after staring at a series of male faces for a long time, a face

that previously appeared gender neutral will look female. Similar aftereffects are very

famous in other domains (look on the web for color adaptation effects and the waterfall

illusion) but it’s still pretty surprising to see such strong adaptation for human faces.

What we are going to do is write a program where subjects adapt for 2 minutes to a set of

male or female images, and then carry out a series of trials where they respond whether a

morph image (morphed between male and female) is perceived as being male or female.

Between each trial will be a 5 second adaptation top-up period. The goal is to measure

how adaptation to male or female images influences your perception of the morphed

images.

You should have a website where you can collect the necessary images. Make a new

folder in your MatlabClass/ folder called ‘Face Experiment’. Then download the images

into that folder. You should have a folder called ‘ male’ , a folder called ‘ female’ and a

folder called ‘ morph’ . In the male and female folders you should have 10 images:

male01, male02 … etc. In the morph folder you should have morphs going from morph00

to morph49.

So let’s think about the structure of the program. You can think of it having 5 parts, as

shown below.

Decide if adapting to m or f

Pre-adapt for X seconds

to series of m or f images

Pre-adapt for X seconds

to series of m or f images

Present test morph image

& get response

Present test morph image

& get response

Repeat X times

Top-up adaptation for X

seconds to series of m or

f images

Top-up adaptation for X

seconds to series of m or

f images

Analyze Data

Chapter 4 - 2 - 4/18/2007

(I’ve put the Analyze Data part in gray since we’re not going to worry about that right

now.)

Each sub-part of this program can also be diagrammed.

Pre-adapt for X seconds to series of m or f images

Load new m or f image

Repeat X times

Display image for X seconds

Number of repeats =

total pre-adapt time

display time per image

total pre-adapt time

display time per image

Whenever you create a program with multiple parts you should make the effort to

diagram it in this way. Trust me – you will end up writing better code and it will save you

time in the long run. Now create a program in your MatlabClass/Face Experiment folder

called FaceAdaptation.m. Now away we go.

We are actually going to start by programming each sub-part of the program as a separate

little chunk.

% Face adaptation experiment

% A program where subjects adapt for 1 minute to

% a set of male or female images.

% then carry out a series of trials where they respond

% whether they perceive a morph image (morphed between male and female)

% as being male or female.

% Between each trial will be a 5 second adaptation

% top-up period.

% Measures how adaptation to male or female images

% influences your perception of morphs.

% written IF Mar 2007

cd( 'C:\WORK\Matlab class & book\MatlabBookOrganization\Spring2007\MatlabClass\Face

addpath(pwd);

addpath( 'morph' );

% adapting variables

Chapter 4 - 3 - 4/18/2007

stim.preadapttime=10; % pre-adapt time (in seconds)

stim.adaptpresentationrate=2; % presentation time (secs) for each adaptation image

stim.numadaptimages=10; % there are 10 images in each adaptation set of images

stim.preadaptstoptimes=stim.adaptpresentationrate:stim.adaptpresentationrate:stim.p

adapttype=input( 'Do you want to adapt to males (m) or females (f)? ... ' , 's' );

if adapttype == 'f'

addpath( 'female' );

adaptstr= 'female' ;

elseif adapttype == 'm'

addpath( 'male' );

adaptstr= 'male' ;

else

disp( 'Error 1. Sorry, response about adapt type is not correct' );

end

pause % waits for a keypress before going to the next stage of the program

%%%%%%%%%%%%%%%%%%%%

% Preadapting interval

%%%%%%%%%%%%%%%%%%%%

adaptcount=1; % keeps track of which adaptation image is being presented

tic % starts a stopwatch

for a=1:stim.preadapttime/stim.adaptpresentationrate

figure(1)

% create a string that matches the filename of the image you want to

% read

if adaptcount<10

imname=[adaptstr, '0' , num2str(adaptcount)];

else

imname=[adaptstr, num2str(adaptcount)];

end

img=imread(imname, 'JPG' );

image(img);

axis off ; axis equal ;

colormap(gray(256));

title( 'pre-adapt' );

drawnow

while toc<stim.preadaptstoptimes(1)

;

end

stim.preadaptstoptimes=stim.preadaptstoptimes(2:end);

% move on to the next image, and keep cycling through the images, so

% adaptcount goes between 1-10

adaptcount=adaptcount+1;

adaptcount=mod(adaptcount, stim.numadaptimages)+1;

end

pause % waits for a keypress before going to the next stage of the program

%%%%%%%%%%%%%%%%%%%%

% Test image

%%%%%%%%%%%%%%%%%%%%

% test variables

ntrials=10;

stim.testindices=linspace(0,49,8);

Chapter 4 - 4 - 4/18/2007

stim.numtestimages=length(stim.testindices);

beep=sin(1:0.5:100);

trialcount=1; % keeps track of which test image is being presented

if stim.testindices(trialcount)<10

imname=[ 'morph0' , num2str(stim.testindices(trialcount))];

else

imname=[ 'morph' , num2str(stim.testindices(trialcount))];

end

img=imread(imname, 'JPG' );

figure(1)

image(img);

title( 'test' )

axis off ; axis equal ;

colormap(gray(256)); drawnow;

% get response

sound(beep);

FlushEvents;

response.char=input( 'Press m for male, f for female' , 's' );

title([ 'response was ' , response.char]);

pause % waits for a keypress before going to the next stage of the program

%%%%%%%%%%%%%%%%%%%%

% Top Up interval

%%%%%%%%%%%%%%%%%%%%

stim.topuptime=6; % seconds for each top up interval

stim.topupstoptimes=stim.adaptpresentationrate:stim.adaptpresentationrate:stim.topu

adaptcount=1;

tic

for a=1:stim.topuptime/stim.adaptpresentationrate

if adaptcount<10

imname=[adaptstr, '0' , num2str(adaptcount)];

else

imname=[adaptstr, num2str(adaptcount)];

end

img=imread(imname, 'JPG' );

image(img);

axis off ; axis equal ;

figure(1)

colormap(gray(256));

title( 'top-up' )

drawnow

while toc<stim.topupstoptimes(1)

;

end

stim.topupstoptimes=stim.topupstoptimes(2:end);

% move on to the next image, and keep cycling through the images, so

% adaptcount goes between 1-10

adaptcount=adaptcount+1;

adaptcount=mod(adaptcount, stim.numadaptimages)+1;

end

pause % waits for a keypress before going to the next stage of the program

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

133

Chapter 4 - 5 - 4/18/2007

%%%%%%%%%%%%%%%%%%%%%

% blank interval

%%%%%%%%%%%%%%%%%%%%%%

stim.shortpause=2;

blankcmap=ones(256, 3);

colormap(blankcmap);

title( 'small pause' );

drawnow;

tic

while toc<stim.shortpause

;

end

134

135

136

137

138

139

140

141

142

143

144

145

The pre-adapt section of the experiment

Line 20-23. Here we have bundled all the parameters we are going to use for the pre-adapt part of the

experiment into a structure called stim. Structures are like little shopping baskets – they allow you to collect

a bunch of variables that are related to each other into a little bundle. The reason you want to do this will be

clearer later. Here we’ve bundled all the stimulus parameters together under the structure stim .

Line 23. Look at stim.preadaptstoptime . It basically is a list of the times at which we want to

move to the next adapting image. We want to move onto the next image every 2 seconds.

Lines 25-34. Here we are checking what response the subject gave. Based on whether the subject

responded ‘m’ or ‘f’, we are setting the path and the name of the string that will be used to pull up jpg image

files. Note that if the subject gives a meaningless response we put in an Error. Good error management will

make your programs much more robust against the random behavior of undergraduate subjects.

Line 41. tic starts a stopwatch. We’ll find out how much time has gone by since we started the stopwatch

using toc

Line 42 . We want to adapt for 10 seconds ( stim.preadapttime ), and we want to present a new

image every 2 seconds ( stim.adaptpresentationrate ). So we will need to present a total of 5

images ( stim.preadapttime/stim.adaptpresentationrate). We’re therefore going

to go through this loop 5 times, each time presenting a face image.

Lines 46-51. Here we are creating a string that represents the filename of the image we want to present on

the screen. We want to present the image with number adaptcount . The first time we run through this

loop, adaptcount will equal 1 (see Line 40).

Line 51. Here we load in the image. The image is saved as a jpg. When loaded in, it becomes a matrix, like

the ones we have been using. But if you whos your workspace you will notice something odd about img -

it’s described as a uint8 rather than a double. The reason for this is that jpg often saves images as unsigned

integers (whole numbers between 0-255) in order to save memory (depending on the particular settings that

were chosen when the jpg image was saved). When Matlab reads in the jpg file it actually reads from the

header of the jpg file that the values inside the file are unsigned integers, and it therefore loads the matrix as

unsigned integers.

> whos

Name Size Bytes Class

a 1x1 8 double array

adaptcount 1x1 8 double array

stim.adaptstr 1x6 12 char array

adapttype 1x1 2 char array

img 540x720 388800 uint8 array

imname 1x8 16 char array

stim 1x1 520 struct array

If you look at img, you will see that the values in img range between 0 and 255.

>max(img)

>min(img)

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: