Sharath Sriram, an associate professor at the RMIT University, Melbourne, recently won the prestigious 2016 3M Eureka Prize for Emerging Leader in Science. Awarded by the Australian Museum, the prizes recognise “excellence in the fields of research & innovation, leadership, science communication and school science.” Sriram, who came close to winning in 2015, finally cinched one this year for work he did with his colleagues on reproducing the way the human brain processes information by developing “an electronic long-term multi-state memory cell.”
Sriram teaches and mentors students in addition to his research. The Wire interviewed him over email about his award-winning research, what got him interested in the field and how he feels about research itself. His answers have been edited for clarity.
How do you describe your research to people outside your field? Is there a go-to explanation?
All the electronics we interact with, such as smartphones and laptops, have memories that store information in the form of 0s and 1s. Each element that stores this ‘0’ or ‘1’ information is a memory cell. If we can go beyond storing just ‘0’ and ‘1’, we take the memory from being digital to analog – capable of packing much more information within the same space. Most significantly, the information stored is going to depend on the information already present – remembering previous state/experience.
How did you get interested in this work?
During the course of my postgraduate education, I really got fascinated by the ability to be able to grow materials and incorporate these into micro or nano-scale devices. I have tried to push the boundaries by taking what a field has considered an issue and trying to turn that into a feature or functionality.
When I think about ‘memory’, I tend to think of computer and human memory as very distinct from each other. While computers seem methodical and logical, human memory is an abstract, highly contextualised and emotional experience to me. How have you come to conceptualise ‘memory’ over the course of your research?
You are right – they are completely distinct. One is managed by programming how information is accessed, while the other is still a mystery in many ways. The main thing we have done is to be able to store multiple information states in a single cell – sort of like a greyscale or analog memory. This can be likened to the difference between a regular light switch (digital memory) to a light with dimmer control (multi-state memory).
The Sydney Morning Herald mentioned the possibility of your research leading to bionic brains, which could be programmed to demonstrate symptoms of diseases such as dementia and help in testing new medications, etc. How far away are we from such an invention? What are the possible limitations and setbacks?
The ability to store multiple states and the fact that the information can relate to previous states relates to experience and cognition (all in an electronic context). A number of the conduction mechanisms that happen in the material that cause this information storage closely resemble conduction in the synapses in the human nervous system. Beyond our team, others have shown the ability to train this category of memory cells (termed memristors) to recognise simple images. This was by combining memory technology and programming. Creating clusters of memory cells that mimic synapse-like functions can allow us to create controllable defects and model medical conditions. Such technology is probably about 10 years away. The main limitation is to bridge the gap between engineering and medicine, to pool collective experience and knowledge to create practical systems.
When we talk about research like yours in the media, we tend to focus on macro impacts such as bionic brains and treating dementia, but when you started this, what was your original research question? Did you know it was going to be a big project or did you discover that over the course of your work?
When starting this project, we knew the long-term impact could be a memory system that mimics the brain. In a simpler engineering context, we felt its components could be the functional hardware elements for artificial intelligence (AI). Combining multi-state memories with all the advances in AI software could unlock new levels of learning and electronic cognition. Towards this, our basic research question was: can we create a stable, durable memory cell that can store multiple information states?
Research itself can be a long process that can feel wearisome or tedious at times. As a researcher yourself and as an advisor to students, how have you coped with such setbacks and what advice do you give your advisees?
I actually find a quote attributed to Winston Churchill a very useful inspiration: “Success is the ability to go from one failure to another with no loss of enthusiasm.” The easiest way to tackle this to admit that in research there would be more failures than success, and to see what we can learn from this. To all students, we keep stressing that a negative result is never a bad thing. If we get a result that we do not expect or predict, that is actually significant and interesting, as explaining why it happened could reveal something new.
What keeps you intellectually motivated when you feel like you’re stuck in a rut?
I have developed the habit of working on many projects at the same time and in diverse fields. This is a challenge but is useful as some project is always showing promise when another gets stuck in a rut. In the current research environment, where metrics matter, the successful projects keep those flowing. It creates the opportunity to give time and resources to the project stuck in a rut and make it work well.