Transcript PPT
Eye movements: a primer Leanne Chukoskie, Ph.D. Two classes of eye movements • Gaze-stabilizing – Vestibulo-ocular reflex (VOR) – Optokinetic Nystagmus (OKN) • Gaze-shifting – Saccades – Smooth pursuit eye movements VOR OKN OKN Gaze-shifting movements for directing retinal specialization Gaze-shifting movements for directing retinal specialization • If you have a fovea, or something like it, you want to be able to aim it. Smooth Pursuit Saccades • Saccades and pursuit are voluntary (vs. OKN and VOR which are stimulus-driven and involuntary), BUT we are thankfully not typically conscious of our saccadic and pursuit eye movements. We can become aware of them and direct them, however. We’ll come back to this point. 3-4 saccades per second! This ends up being the most common movement you’ll make in your life! Let’s study it! • The saccade itself is fast, and its velocity is dependent on the amplitude of the saccade– that is, bigger saccades are faster. This is relationship is called the ‘main sequence’. Main sequence We don’t see during saccades • We are effectively blind during saccades. • Why? Two possible reasons: 1) typical speeds of movement during saccade lies outside the visual modulation transfer function– we just don’t see things moving that fast. 2) an active process believed to be in the visual thalamus “gates” visual processing during a saccade. • Perisaccadic mislocalization of a target also occurs, suggesting potential mechanisms. Saccade Circuitry How do we measure eye movements? • Electrooculography – Small voltages can be measured from around the eyes which vary with position. Easy and cheap but subject to drive and high levels of inaccuracy. • Video-based methods are most popular today – Using computer vision tools to identify the center of the pupil and also often the corneal reflection. Use these in algorithms to estimate gaze position but importantly involve IR-light sensors for greater robustness. • Scleral search coil method – Observer wears a large contact lens with a wire embedded in it and sits inside a chamber of magnetic field. When the coil of wire moves, it induces a voltage Electrooculography Small voltages can be measured from around the eyes which vary with position. Easy and cheap but subject to drive and high levels of inaccuracy. Video-based methods are most popular today • Using computer vision tools to identify the center of the pupil and also often the corneal reflection. Use these in algorithms to estimate gaze position but importantly involve IR-light sensors for greater robustness. Scleral search coil method Observer wears a large contact lens with a wire embedded in it and sits inside a chamber of magnetic field. When the coil of wire moves, it induces a voltage. This is really accurate but painful and inconvenient so it isn’t used much any more. What do we typically measure? • Most of the time in simple saccade tasks we look at accuracy and latency • For scene-based tasks, we also want to know what the observer looked at and it is less clear how to calculate accuracy per se. Saccades Need better methods • Even though we’ve been studying eye movements since the early 1900s, we still struggle with how to analyze sequences of gaze. • Most methods involve chunking a scene and then comparing sequences of what was fixated Saccade endpoint choice • Driven by a combination of low level salience and higher level cognitive factors. These factors also impact the amount of time people fixate. Passive vs. active paradigms • Passive: I can measure what you looked at on the screen • Active: I can use your gaze movements to actively CHANGE what is presented on the screen. This has been used to study the “preview benefit” in reading (Rayner). We are using these “gazecontingent” methods to make games to train faster and more accurate gaze re-direction. What can we do with these measurements of behavior? • Gaze behavior changes across development. • Disorders, such as schizophrenia, autism spectrum disorder • Alzheimer’s Disease– interesting potential way of identifying early risk– look at how people scan a series of images that were presented twice. How individuals scan the second presentation can tell you about what they remember (Stuart Zola)