Lecture 26: Eye Tracking

Transcription

1 Lecture 26: Eye Tracking Inf1-Introduction to Cognitive Science Diego Frassinelli March 21, 2013

2 Experiments at the University of Edinburgh Student and Graduate Employment (SAGE): Researchers (normally desperate PhD students) are looking for participants for their experiments You can learn new interesting methodologies You can gain some money In your 3rd year you can volunteer for helping Eye Tracking 2/14

3 An Eye is Not a Camera Vision is a process that produces from images of the external world a description that is useful to the viewer and not cluttered with irrelevant information. (David Marr, Vision, 1982) Back to lecture 4 The eye is not a passive recorder Vision involves many layers of active interpretation and processing A process that maps one representation to a different one Eye Tracking 3/14

4 Eye Tracking The Eye-Mind Hypothesis (Just & Carpenter, 1980) Where participants are looking indicates what they are processing. How long they are looking at indicates how much processing effort is needed. An eye-tracker makes possible to record the eye-movements of participants while they are performing a cognitive task Based on a slide by Frank Keller. Eye Tracking 4/14

5 Some History Louis Emile Javal (1879) he attached a microphone to the closed eyelid of a person when the person was reading (with the other eye) the microphone was recording the noise produced by the cornea colliding with the microphone Edmond Delabarre (1898) he put a plaster cap in his eye ( sufficiently cocainised ) the cap had a hole for the pupil the cap was wired to a lever which drew horizontal lines on a panel when the eye was moving during reading Based on a slide by Tobii eye-tracking research. Eye Tracking 5/14

6 Nowadays More freedom and more natural information But also more noisy and less accurate data Eye Tracking 6/14

7 How does it work Find a mapping between the eye position and the image gazed A camera records the eye movements projecting some infra-red light against the eye of the subject The cornea and the pupil reflect infra-red light: easier to recognise and to track their movements (no heavy image recognition) Calibration process: providing some examples of the area fixed and the reflections produced by the cornea and the pupil An algorithm hew window should get opened when a link leads out of the current docuas to superimpose the fixations to the image recorded (gaze estimation) Eye Tracking 7/14

8 How does it work (contd.) The eye tracker records two eye movement events (but not only!): Fixations: collection of most of the visual information ( ms) Saccades: a rapid movement from one fixation to the other (30-80 ms). They are the fastest body movements. We are blind during most of them The experimenter decides what a fixation is Eye Tracking 8/14

9 Different Scenarios Nowadays, eye-trackers are used in different fields: Scene Perception (Playing Cards) Web Design (Ikea Website) Marketing analyses (Supermarket) Sport studies (Ronaldo) Eye Tracking 9/14

10 Different Scenarios - Google Glasses We created Glass so you can interact with the virtual world without distracting you from the real world. We don t want technology to get in the way. Google designer Isabelle Olsson Still a prototype It adds another layer to reality Multitasking does not exist How does this information affect perception? Google Glasses Project Eye Tracking 10/14

11 Eye-tracking and Cognitive Science Can you think of other applications of eye-tracking? Come up with cases in which recording eye-movements is useful to study: Language Processing Visual Cognition Memory Cognitive Impairment Based on a slide by Frank Keller. Eye Tracking 11/14

12 Designing your own experiment Clearly formulating the research question of your experiment is the first step for producing a good design The null hypothesis (H0): no effect is expected between two or more conditions An experiment is aimed to reject H0 supporting the alternative hypothesis (H1) Eye Tracking 12/14

13 Designing your own experiment - Independent Variables Independent variables: the conditions manipulated by the experimenter increasing the amount of IV requires a higher number of subjects: You have to find them You have to pay them You have to spend time collecting the data The equipments and the lab are not always available and they cost money Eye Tracking 13/14

14 Designing your own experiment - Dependent Variables Dependent Variables: the outcome variables not manipulated by the experimenter: Number of fixations towards a specific target Reaction times: time required to perform an action Error Rates: number of mistakes occurred The question is directly related to the technology we use Eye Tracking 14/14