Author: Erika H. Siegel, PhD
The HP Omnicept VR platform is a ground-breaking collaboration between HP’s VR software team and scientists at HP labs. Driven by research in neuroscience, psychology, and affective computing, Omnicept VR combines physiological sensing with state-of-the-art machine learning to create targeted, human-centered, AI solutions. HP Omnicept VR will help individuals and teams better understand their users, predict behavior, and optimize performance.
Our models are developed and rigorously tested by a small team of senior scientists with expertise in machine learning, signal processing, user experience, design, and psychophysiology. We build solutions from the ground up based on well-established and well-understood science. We collect (and protect) our own data and develop models to predict outcomes tailored to solutions.
Our first Omnicept VR software release includes an SDK (software developer kit) for VR developers with an AI inference engine that can make real time predictions of an individual’s mental effort while they complete a task. VR training and learning solutions are in high demand and our SDK offers a way to incorporate real-time predictions of how effortful a task is for users. Research suggests that biometric indicators increase proportionally with mental effort[i] so our models incorporate information from the heart (PPG, pulse plethysmography), the pupil (pupillometry), looking behavior (eye tracking), and tracking information from the HMD and controllers.
What is mental effort and why does it matter?
In scientific terms, the amount of mental effort required to perform a task or learn something new is called cognitive load and has been studied by researchers interested in learning, memory, and performance for more than century[ii]. Every person has their own information processing capacity (also called working memory capacity or short-term memory) and it is fixed, limited, and varies from person to person. Information processing capacity is like the amount of food a person can hold in their mouth at one time. Sure, maybe it is possible to fit 44 marshmallows in your mouth at once, but is it comfortable? Probably not. Is it an optimal way to eat? Definitely not. What about a single sesame seed? Would you even bother chewing and swallowing it? Perhaps, but it would be a waste of energy.
If we think of information processing capacity the same way, then cognitive load (CL) is like the number of marshmallows in your mouth at a given moment. If you have too much, you can’t think; too little and it’s not worth the effort. Figure 1 (below) demonstrates the relationship between task performance and cognitive load. When cognitive load is very low, performance is often poor because the task is so easy it’s boring. When cognitive load is too high, performance is weak because users are mentally overloaded. However, when performance and cognitive load are both at optimal levels, users are in a type of Goldilocks zone (sometimes called a “flow state”). They are engaged, energized, and able to complete the task in comfort. Psychological research suggests that optimal cognitive load leads to better information retention, less exhaustion, and greater enjoyment of the task.
Cognitive load is also an important indicator of expertise. The amount of cognitive load required to successfully perform a task typically decreases with training. With repeated trainings, cognitive load offers novel insights about a users’ learning trajectory. As we see in Figure 2, novices typically experience high cognitive load combined with low performance because they are acquiring a new skill. As they develop proficiency, their performance improves, but the task remains effortful. As users become experts they can achieve consistently high performance without being overloaded.
Omnicept Enhanced Training
With the Omnicept SDK your insights get a level up as you peek into the minds of users in real time. You will now have the ability to personalize training like never before. When combined with metrics of performance, Omnicept allows you to shorten training with confidence or dynamically change the level of difficulty to meet trainees where they are. Real time cognitive load estimates provide the opportunity to deliver targeted interventions to help your users when they need it most.
Omnicept Enhanced Design
Omnicept can offer important insights for industrial design and product development teams. Cognitive load levels can index of how comfortable products and designs are to use. A/B testing is enhanced with the addition of CL, helping teams interpret qualitative feedback and make more informed decisions about what to build. Industrial designers can use CL levels as a way to understand areas of confusion in simulated design environments.
The HP Omnicept platform is extensible and will continue to evolve as new technologies and partners emerge. It’s designed to work with XR engines like Unity and Unreal. And, it’s secure to develop and deploy anytime, anywhere. What can Omnicept do for you?
[i] Kahneman, D., Tursky, B., Shapiro, D., & Crider, A. 1969. “Pupillary, Heart Rate, and Skin Resistance Changes during a Mental Task.” Journal of Experimental Psychology 79 (1): 164–67
[ii] Taylor, F.W. (1913). The principles of scientific management. New York: Harper and Bros.