ARTEMIS

ARTEMIS

Recognize the Undiscovered

 

Screenshot of the ARTEMIS virtual reality demo using Feynman diagrams.

In pursuit of scientific discovery, especially in fields which rely on quantitative approaches and mathematical models, one way to broadly define the “discovery cycle” is as a sequence containing three key phases: (1) conceptualization, (2) calculation, and (3) interpretation.

Initially, in the conceptualization phase researchers ask questions, generate ideas, and shape those ideas into mathematical models that can be tested.

Next, in the calculation phase researchers use those models to extract numerical predictions and/or to attempt to fit observed data from experiments designed to explore, confirm, or discover various phenomena.

Finally, in the interpretation phase researchers review the numerical outputs of their models and calculations to try and answer their initial questions, to gain insight and understanding, as well as to direct their focus in new directions.

Of the three phases outlined in this cycle, the final two–calculation and interpretation–have received significant attention, especially in the last few decades as computational power has grown and tools like machine learning and scientific visualization have matured.  The utility of software and computational tools to automate, i.e., replace, human effort, and/or significantly augment, i.e., support, human effort, with respect to scientific discovery, has been driven forward by a vast community of dedicated and capable researchers.

However, the role of computational tools in the phase of conceptualization–of coming up with new ideas and questions related to scientific discovery–has remained nascent.  In part, because this is a challenge that lends itself more to augmenting human thought, a delightfully messy and complicated process, rather than replacing human calculation, a well-defined tangible sequence.  The “Eureka” moment remains too elusive to code in an algorithm.

But a vast body of evidence-based research exists on many of the key elements of how to foster, focus, and facilitate the necessary thinking that goes into conceptualization, especially ways to engender the necessary blending of qualitative and quantitative skills that make scientific discovery unique.  Many of these skills are best supported in ways that include sensory-motor experience as well as traditional math and language skills.

Screenshot: making changes to the physics content of a Feynman diagram in virtual reality using ARTEMIS.

In recent years, the rapid commercialization of virtual reality (VR) and augmented reality (AR) technology has opened up the possibility of finally designing such a computational tool.

My research group has undertaken the task of building a first-of-its-kind VR tool designed to support the conceptualization process and its role in scientific discovery.  In Summer 2017, two students at the University of Manchester built a proof-of-concept software demo allowing users to manipulate in VR, through hand gestures captured by a  camera, virtual models of something known as Feynman particle diagrams from particle physics–diagrams representing both the mathematics and intuition behind how fundamental particles, like electrons and neutrinos, interact.  This tool was named ARTEMIS: Artificial Reality Tool for the Enhancement and Manipulation of Insight in Science.

Above: another screenshot of ARTEMIS. Below: the ARTEMIS demo “particle grid” where users can pick up elements to build their own Feynman diagram by hand.

Theoretical particle physics is notorious for its level of difficulty, in regards to scientific discovery, partly because of its often abstract and counter-intuitive nature.  My own background is in the subtopic of theoretical and applied neutrino physics, which spans both particle physics and nuclear physics.  Neutrinos are fundamental particles, very weakly interacting (they can pass right through roughly one solid lightyear of lead without interacting), which are pervasive throughout the universe.  They appear repeatedly in discussions of physics discovery-related questions, such as: Why is there more matter than antimatter in the universe?  What lies beyond the Standard Model of Particle Physics that we don’t yet know about?  What is dark matter?

As such, neutrinos make an excellent testing ground for any tool designed to help us think of new and incisive questions about what lies beyond the physics we already know, and to come up with answers to those questions that already keep physicists awake at night.

In 2018-2019, my research group looks forward to evolving ARTEMIS from a proof-of-concept demo to a working prototype able to tackle rudimentary scientific discovery questions in theoretical neutrino physics.

 

(For more on my original thoughts behind the ARTEMIS virtual reality tool, being designed to augment our ability to conceive of new scientific discoveries, see my blog entry here.)