Everything you always wanted to know about eye tracking (Part 3)

Written by on October 31, 2016 in Features with 0 Comments

Get the most from eye tracking

visible_light_eye-tracking_algorithmWhy combine eye tracking with other biometric sensors?

Take up again the milk bottle situation in (Part 2). Although you looked straight at the bottle, you didn‘t realize it was standing in front of you. Based on eye tracking alone we would have argued that, since your gaze was directed toward the item, you must have seen it.

The most intuitive way to validate our assumption might be to just ask you – “did you see the milk bottle?”. Primarily due to the relatively low costs, surveys are in fact a very common research tool to consolidate gaze data with self-reports of personal feelings, thoughts, and attitudes. Of course, the power of self-reports is somewhat limited when it comes to the disclosure of sensitive personal information (alcohol, drugs, sexual behavior etc.). Also, when working with self-reports, keep in mind that any delay between action and recollection introduces artifacts – asking immediately after closing the fridge might yield a different response (“nope, I didn‘t see it!”) compared to one week later (“…um, I don‘t remember!”).

Take your study from ordinary to extraordinary

We could have used camera-based facial expression analysis to monitor your emotional valence while staring into the fridge, for example. A confused expression while scanning all items in the fridge followed by a sad expression when closing the door would have told us immediately that you obviously didn‘t realize the milk bottle was standing within reach. On the other hand, a slight smirk would have indicated that your search had been successful. Further, we could have evaluated the amount of your emotional arousal and stress levels based on the changes in skin conductance (GSR, EDA) or your heart rate (ECG).

On top, we could have used EEG to capture your cognitive and motivational state as it is the ideal tool to identify fluctuations in workload (“have I looked everywhere?”), engagement (“jeez, I have to find this bottle!”), and drowsiness levels (“are you kidding me? I need coffee asap!”).

Admittedly, our example pictures a simplification of the interpretation of physiological data. In most research scenarios you will have to consider and control for many more factors that might have a significant impact on the parameters you aim to select.

Each sensor reveals a specific aspect of human cognition, emotion, and behavior. Depending on your individual research question, consider to combine eye tracking with two or more additional biosensors in order to gain meaningful insights into the spatio-temporal dynamics of attention, emotion, and motivation.

What‘s the gain?

Sensor combinations with eye tracking

The true power of eye tracking unfolds as it is combined with other sources of data to measure complex dependent variables.

These 5 biometric sensors are a perfect complement to eye tracking. Which metrics can be extracted from the different systems?

Have a look.

ECG & PPG: Electrocardiography (ECG) and Photoplethysmography (PPG) allow for recording of heart rate (HR), or pulse. Get insights into respondents’ physical state, anxiety and stress levels (arousal), and how changes in physiological state relate to their actions and decisions.

Facial Expression Analysis: Facial expression analysis is a non-intrusive method to assess both emotions (subconscious reactions prior to feelings – typically small movements in face muscles) and feelings (conscious reactions occurring after emotions – typically more visible muscle movements). While facial expressions can measure the presence (valence) of an emotion/feeling, they can´t measure the power of that emotion/feeling (arousal).

EEG: Electroencephalography is a neuroimaging technique measuring electrical activity on the scalp. EEG tells which parts of the brain are active during task performance or stimulus exposure. Analyze brain dynamics of engagement (arousal), motivation, frustration, cognitive workload and other metrics associated with stimulus processing, action preparation, and execution. EEG provides the quickest response of all biometrics sensors.

GSR (EDA): Galvanic skin response (or electrodermal activity) monitors „emotional” sweat secretion on hands or feet. Skin conductance offers insights into the respondents’ subconscious arousal when being confronted with emotionally loaded stimulus material.

EMG: Electromyographic sensors monitor the electric energy generated by bodily movements (e.g., of the face, hands or fingers). Use EMG to monitor muscular responses to any type of stimulus material to extract even subtle activation patterns associated with emotional expressions (facial EMG) or consciously controlled hand/finger movements.

Best practices in eye tracking

Frankly, failure or complications in studies most often occur due to small mistakes that could have easily been avoided. Often they happen because researchers and staff just didn‘t know about the basic knacks to avoid running into issues.

Conducting an eye tracking study involves juggling with a lot of moving parts. A complex experimental design, new respondents, different technologies, different hardware pieces, different operators. Let that sink in for a few moments. On top, you‘re under pressure to achieve a great study outcome. Unless you‘re an old hand in research, it admittedly can get quite challenging at times. We’ve all been there.

Don‘t worry, we‘ve got your back. The following are our 6 safe bets for a smooth lab experience in eye tracking research.

eye-tracking-best-practices

 

1. Environment and lighting conditions
Have a dedicated space for running your study. Find an isolated room that is not used by others so you can keep your experimental setup as constant as possible. Make sure to place all system components on a table that doesn‘t wobble or shift.
For eye tracking, lighting conditions are essential. Avoid direct sunlight coming through windows (close the blinds!) as sunlight contains infrared light that will impact the quality of the eye tracking measurements.

Avoid brightly lit rooms (no overhead light). Ideally, use ambient light. Ensure to keep the physical environment constant.
If your study is designed to run several hours, split the experiment into two equally long sessions and calibrate the system for each session separately. Also, be aware that long experiments might cause dry eyes, resulting in impeding drift. Try to keep noise from the surrounding environment (rooms, corridors, streets) at a minimum as it most likely will distract the respondent and affect measurement accuracy.

2. Work with dual screen configuration
Work with a dual screen configuration – one „respondent screen“ for stimulus presentation (which ideally remains black until stimulus material pops up) and one „operator screen“ (which the respondent should not be able to see) to control the experiment and monitor data acquisition. A dual screen setup allows you to detect any issues with the equipment during the experiment.

3. Clean your computer before getting started

  • Clean your computer from things you don‘t need
  • Disable anything securing your computer, e.g. make sure to turn off your antivirus software so it doesn‘t pop up during the experiment and uses CPU resources
  • If not needed, disconnect your computer from the internet during data collection
  • Disable your screen saver
  • Disable any pop ups that could disturb the experiment

4. Ensure all people involved are properly trained
It is essential that people who start at the lab are trained on the systems used so that they have a certain level of knowledge that allows them to run a study smoothly. Generally, trainings are important for any kind of position in the lab. Having to train people during the testing process is a disadvantage and usually requires more time and effort than needed with solid training beforehand.

5. Always use protocols
Always have protocols! Anything that you need to instruct somebody on or any documentation that is associated with setting up and/or running a study at the lab is the most important thing to have. Try to have templates for every step of the research process. Literally. Don‘t underestimate the importance of documentation in institutions such as a university. It‘s common practice for research assistants to switch labs after a certain amount of time – protocols are true lifesavers as they keep track of anything from management to study execution and therefore make sure that new lab members can jump right in and are able to perform on the fly in line with lab routines.

6. Simplify your technology setup
Chances are you need a couple of different biometric sensors to run your study. To make sure that they interact well and are compatible with each other, use as few vendors as possible for both hard- and software. In the ideal setup everything would be integrated in one single software. Having to switch between different operating systems or different computers can cause difficulties. Keep in mind that it‘s easier to train lab members on a single software than on multiple. The equation is simple: Having a single software platform decreases the amount of training needed, simplifies the setup and takes out the risk of human error. Also, in case of problems and support issues it is more convenient to deal with one vendor and have a direct contact person than to be pushed around between vendors because nobody feels responsible.

N.B. This is an excerpt of our free white paper “Eye Tracking – The Pocket Guide”. To get the full guide in an easier to read PDF file format click here.

Equipment 101: Hardware

At this point you might figure that all eye tracking systems are pretty much the same. Their only job is to track where people are looking, so what‘s left to vary? Actually, quite a lot. Eye tracking is on the rise, and to keep pace with demand, new systems are shooting up like mushrooms. Amidst all manufacturer specifications it can be quite hard to keep the overview and evaluate which eye tracker is right for you.

eye-tracking-hardware

 

Which eye tracker is right for you?

  1. Will your respondents be seated in front of a computer during the session? Go for a screen-based eye tracker. Do your respondents need to move freely in a natural setting or virtual reality? Choose a head-mounted system that allows for head and body mobility.
  2. Make sure the eye tracker you purchase meets the specifications required to answer your research objectives. Have a look at the key questions below that can help you find a suitable eye tracker.
  3. We have that covered already, but here it goes again: Even though less expensive, you rather stay away from eye trackers using ordinary webcams. Yes, we know the temptation of a good bargain – however, when it comes to eye trackers it‘s absolutely worth spending a bit extra money if you‘re aiming for high measurement accuracy.

Here are a few more questions to ask to get the ideal performance picture:

Measurement precision: Measured in degrees. Standard is about 0.5 degree. Low end hardware start around 1.0 degree, medium 0.5 degree, high end down to 0.1 degree or less (bite bar).

Sampling rate: How many times per second is the eye position measured? Typical value range is 30-60 Hz. Special research equipment samples at around 120-1000+ Hz.

Trackability: How much of the population can be tracked? The best systems track around 95% of the population, low end systems less.

Headbox size: To what extend is the respondent allowed to move in relative distance to the eye tracker? A good system will typically allow to move around 11 inches in each direction.

Recapture rate: How fast does the eye tracker detect the eye position after the eyes have been out of sight for a moment (e.g. during a blink)?

Integrated or standalone: Is the eye tracking hardware integrated into the monitor frame? Standalone eye trackers are more flexible, but typically a bit more complex to set up.

Does your provider offer support? Thanks to plug and play, you basically can run your eye tracker out of the box (that pertains to most eye trackers, at least). To get started, however, a live training is helpful to learn the ropes. Even along the way, a little expert advice oftentimes come in handy. Does your provider offer that kind of support? What about online support? And how long does it take them to reply when you need it most? Seriously, good support is gold par value.

Eye tracking software

Hardware is only half the battle – finding the right software solution.

Of course, hardware is only half the battle. Before you can kick off your eye tracking research, you definitely need to think about which recording and data analysis software to use. Usually, separate software is required for data acquisition and data processing. Although some manufacturers offer integrated solutions, you will most likely have to export the raw data to a dedicated analysis software for data inspection and further processing. So which eye tracking software solution is the one you need?

What are the usual struggles with eye tracking software?

Struggle 1: Eye tracking software either records or analyzes: Usually, separate software is necessary for data recording and data processing. Despite automated procedures, proper data handling requires careful manual checks along the way. This recommended checking procedure is time-consuming and prone to error.

Struggle 2: Eye tracking software is bound to specific eye trackers: Typically, eye tracking soft- and hardware are paired. One software is only compatible with one specific eye tracker, so if you want to mix and match devices or software even within one brand, you soon will hit the brick wall. Also, be aware that you need a separate software for remote and mobile eye trackers.

Not only does operating several software solutions require expert training beforehand, it might even prevent you from switching from one system to another. Worst case scenario? Your lab will stick to outdated trackers and programs even though the latest generation of devices and software might offer improved usability and extended functionality.

Struggle 3: Eye tracking software is limited to certain stimulus categories: Usually, eye tracking systems don´t allow the recording of eye movements in various experimental conditions. You will have to use one software for static images and videos, a different software for websites and screen captures, and yet another software for scenes or mobile tracking. What if there was one unified software solution for your biometric research? Go to iMotions eye tracking.

Struggle 4: Eye tracking software can be complex to use: Eye tracking software can be quite complex to use. You have to be familiar with all relevant software controlled settings for eye tracker sampling rate, calibration, gaze or saccade/fixation detection etc. In the analytic framework, you have to know how to generate heat maps, select Areas of Interest (AOIs) or place markers. Statistical knowledge is recommended to analyze and interpret the final results.

Struggle 5: Eye tracking software rarely supports different biometric sensors: Eye tracking software often is just that – it tracks eyes but rarely connects to other biometric sensors to monitor emotional arousal and valence.

What does that imply exactly? You will need to use different recording software for your multimodal research. As you most likely will have to set up each system individually, a considerable amount of technical skills is required. If you happen to be a tech whizz, you´re probably on the safe side. If not, you might be running into serious issues even before getting started. Also, be aware that you have to make sure that the different data streams are synchronized. Only then you can analyze how eye tracking, EEG, or GSR relate to each other.

What should a picture-perfect eye tracking software hold in store for you?

Ideally, your eye tracking software

  • connects to different eye tracking devices (mix and match, remember?)
  • is scalable correspondent to your research needs: It allows you to conveniently add other biometric sensors that capture cognitive, emotional or physiological processes
  • accommodates both data recording and data analysis
  • tracks various stimulus categories: Screen-based stimuli (video, images, websites, screen capture, mobile devices), in real-life environments (mobile eye tracking) as well as survey stimuli for self-reports
  • growths with you: It can be used equally by novices that are just getting started with eye tracking and expert users that know the ropes

Tags: , ,

Guest Writer

About the Author

About the Author: This article has been supplied by one of our many guest writers that have specialist interest in the topic featured. .

Subscribe

If you enjoyed this article, subscribe now to receive more just like it.

Subscribe via RSS Feed

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Top
%d bloggers like this: