Language
cover_photo

How to know the suspect are telling you the truth

a recognition psychology experiment implement by using unity

What is Guilty Knowledge Test ?

The Guilty Knowledge Test (GKT), also known as the Concealed Information Test (CIT), is a psychological assessment technique used to determine whether a person has knowledge of crime-relevant information that only the perpetrator would know. Unlike traditional polygraph tests, which attempt to detect lies based on physiological responses to direct questions about the suspect's involvement in a crime, the GKT focuses on identifying the recognition of specific details of an event or activity by measuring physiological responses to multiple-choice questions, each containing one option that is relevant to the crime (the "probe") and several neutral, irrelevant options.

please visit the lab home for more information

tutorial


the following is the tutorial part to explain how to reproduce this experiment

About

This Experiment is an implementation of guilty knowledge test (GKT) . it utilize VR (virtual reality) and pupil lab (Eye Tracker) to let experimenter know where is subject looking at but without the subject's awrareness.

Third Party Document

Table of content

Setting Environment


Add HTC Add-on to your vive

  1. Step 1 : Remove head strap.
  1. Step 2 : Extend depth adjustment.
  1. Step 3 : TAKE HTC Vive apart a bit more so we can show a close up view.
  1. Step 4 : Closeup of the add-on engagement with the lens holder.
  1. Step 5 : Resemble it
  2. Next install the htc vive sdk

HTC vive installation

you can follow up this tutorial or just open the environment setting at SteamVR

HTC Vive setting

please take a look at official tutorial from VIVE

pupil lab capture

Because most of core function of this program is based on the Pupil capture service you will need to open pupil capture and make sure you have already correctly connect HMD Add-on directly to computer.

View-Preview


Title

title_preview (2).jpg

Checking Device

checking_preview.jpg

Before step into experiment, there something you need to prepare In this section you will need to check if all the device has settele down.

  1. Step1 : finished the HTC Vive environment setting
  2. Step2 : open the pupil capture software to get connect with.

if you haven't finish the environment setting you can back to the 『Setting Environemtn』section

mainmenu_preview.jpg

Setting

setting_preview (1).jpg

Use Custom Setting

In this page you can custom the experiment setting including almost every detail . If you broke the setting and don't remember the original version, don't worry there are four features you might want to use

  • save : save current setting to general setting
  • save as : save current as a new setting
  • boxcutter03a : bottom-right(3rd)
  • broom : bottom-left(4th)

How to set custom visual target

first you need to know is the name is matter, each folder means each trial of experiment each image have a position in visual target

  1. Folder name would affect order of experiment visual_path (1).jpg Please Remember the folder name should start with "Trial" ,The computer only reconize the folder which name is start with "Trial" !!! these will determine the order of visual target set that subject see
  2. Image is sort in alphabetical order The image name is also important. Because the position of visual target that subject see in the vr is depend on the order of image img_path.jpg take example : the mapping positions of Assets/StreamingAssets/Visual_target/Trial1 is like below img_sort.jpg
  • axe : top-left (1st)
  • bastinbrush : top-right (2nd)
  • boxcutter03a : bottom-right(3rd)
  • broom : bottom-left(4th)

Key point :

  1. the order of folder : In alphabetical order.
  2. the image name influence order and the order influence image position
  3. Currently only support "png" format image

I'll develop more image format in the future plese look closely if there is any error in the Visual Target Preview it will not record if the image number is incorrect

Experiment

Experiment_preview.jpg

  1. page intro :

At this page you can check the time sapn with (time bar) and the view what the subject see also the eye frame visualizer.

  1. Weak Eye and Strong Eye :

In this preview the left eye is weak eye and right eye is strong eye. Weak Eye has four visual target and will automatically start fade in when FadeIn Stage start , Strong Eye has only one video which is mondrian video (produced by this project have a look if you need some mondrian video sources)

the visual target image source can change by modifing the visual target path at setting page the mondiran video source can also change by modifying the mondiran path at setting page both can be found at StreammingAssets path

  1. Experiment Stage :
  • PupilLab Calibration (only execute at very first time)
  • Data Optimize Calibration : this stage will collect the gaze data that subject gaze calibration point
  • Start Delay : the stage that before fadin stage
  • FadeIn
  • End Delay
  1. Record Data Structure as long as you click start experiment button , the system will automatically generate a experiment setting . feel free if you want to cutomize every setting for your subject

Devlopment


Program overview

the program can be split into three parts

  1. GKT-Experiment : control the experiment process
  2. GameManager : Control the UI Event and the interface between GKT-Experiment and GUI Page
  3. EyeTracker :

Technical detail


Experinment Setting : (JSON)

  • gapTime (3): the start delay time before fade-in process
  • maxTime (10): the time span of fade-in process
  • delayTime (1): the end delay time before end this trial
  • maxAlpha (0.5): the maximum alpha value that subject would see , you can see the preview on the right side
  • mode (0): choose which eye is subject's weak eye , mode 0 is the right side ande mode 1 is the left side
  • mondrian video path (./Assets/StreamingAssets/Mondrian): the folder path that mondrian videos locate
  • visual target (./Assets/StreamingAssets/Visual_target): the folder path that visual_target "folders" locate
  • record path (Desktop): the folder that will auto saving every experiment record
  • record name (Subject_1): the name of record folder , the index of the end would auto increment if the system detect there exist a previous record
  • weak eye : the eye that system would auto track and records, this eye will see the visual target
  • strong eye : this eye will see the mondrian video
  • Calibration Data Format(List, Class : Square Target)

Experiment Record : (JSON)

  • index : the index of experiment
  • top-left : the top-left image name
  • top-right : the top-right image name
  • bottom-left : the bottom-right image name
  • bottom-right : the bottom-right image name
  • finishTime : The Recording time span when subject's weak eye saw the visual targets
  • finalAlpha : The alpha value when subject saw the visual target

Gaze Data (CSV):

  • confidence : the posibility that pupil lab think it is correct data
  • GazeMode : which eye is been record, 0 means right eye , 1 means left eye
  • gazePoint3d : the mapping data from eye ball to visual target
  • standardCalibrationPoint : the position of calibration point in 3d space
  • pupilTimeStamp

Pupil Export Data :

Check the pupil document link

Buy me a coffee