Cognitive Load and how researchers are estimating it?

Source: http://en.people.cn/n3/2017/0321/c90000-9193030.html

Out of all the things in the world, why would someone write about cognitive load? This is what I am thinking exactly, right now. Why not? Don’t you feel overloaded with tasks? As a researcher, I was curious is there any parameter to detect? Not a trivial scale of 1 to 10. Actual data. Then, I came across this term. Cognitive Load Estimation and started digging more.

First, What is cognitive load estimation?

“Cognitive load is the amount of working memory being used, like the RAM in your computer, only for the human brain.”

Oh, yes!! sometimes I do sometimes feel like I am overloading. I am listening, please elaborate.

So, there are basically three types of cognitive load.

Source: https://mcdreeamiemusings.com/blog/2019/10/15/the-good-the-bad-and-the-can-be-ugly-the-three-parts-of-cognitive-load

• The first is the “intrinsic load” the load that is associated with the task at hand
• The second is the “extraneous load,” the load imposed by the format of the stimulus (the User Interface of the experiment)
• The third is the ”Germane,” which is the effort made to process the stimulus

This is the reason we need better UI/UX designers. To reduce the extraneous load. “Tik-Tok” nailed it. Germane load depends on the task and is very subjective. One can feel cognitively challenged when they do simulation(like me), whereas other can do it easily.

ok. But how do I measure it?

Apparently, NASA thought way ahead, for their astronauts. Astronauts have to deal with some difficult and challenging tasks on daily basis. They wanted to have some kind of test to know if they are cognitively fit for the duration of the journey. They came up with, the NASA task load index(TLX), a set of questionnaires developed for estimating human mental workload

Basically, they would hand a set of questions after a said task is done and judge on the basis of 6 parameters. These are mental demand, physical demand, own perception of performance, temporal demand, effort, and frustration.

But these are back-dated methods. I need to know “now”. Here, comes iot in its full glory. EEG, eye-tracking and smart wearables are the most commonly used now. Let’s have a look into how it is done.

First you need to design a stimulus and find some willing volunteers. The stimulus are made of three types Easy, Medium and Hard tasks. The most commonly used is the N-back tests.

N-back tests are where the subject is presented with a string of numbers. like 40,79,43,59,70….after every 1 no. the subject have to repeat the first number. Like for 1-back test after 43, you have to say 40. After 70 one have to say 43. For 2-back, after 59 one has to say 40. This way increase the difficulty level.

How do we measure it now then?

Eye-tracking

Here mainly 4 parameters are tracked.

Eye based Indicators for cognitive Load

• Fixation: The constant maintenance of eye gaze in a particular location is defined as visual fixation.
• Saccades : A saccadic movement is an instantaneous change in the eye between two fixation points.
• Blinks : Blinking is a semi-voluntary activity due to the eyelid’s protractors and retractor muscles.
• Pupil : The variation in pupil size can be due to most factors like age, humidity, and changes in lighting.

Ok. What about the EEG and smart wearables?

Source: https://www.cartoonstock.com/directory/w/wearable_technologies.asp

Here, the data is collected through 31/64 electrodes(EEG) and smart watch. But, the steps are very similar.

• Data Collection : smart watches connected over mobiles phones(R-R (or inter-beat) intervals, galvanic skin response (GSR), heart rate (HR), skin temperature (ST), barometer data, accelerometer and UV index data), 32/64 electrodes headset(EEG).
• Data Preprocessing : After this the data is either standardized or min-max normalization, removal of noise and bias.
• Feature Extraction : Handcrafted or Automatic time-series feature extraction .
• Feature Selection: sequential backward floating search, Gini impurity ,and maximal information coefficient.PCA, ICA are also applied.
• Classifier: svm,KNN,Random Forest, logistic regression, and CNN are commonly used.

Cognitive load is an important use-case. Employers can use it to understand the mental condition of the employees. In military, these becomes crucial as a part of overall fitness. As we move more into this decade, we will see more people doing overwork, having 2–3 jobs at a time. Always running, hustling for just a lil bit more. Maybe if this parameter is widely used, people would stop for a while. Take a deep breath and then work to death.