Unlocking the future: mimicking the human brain for information processing

Meet Matthew: a trailblazer in the realm of neural networks and artificial intelligence, whose journey from Zurich to Groningen in 2023 has embodied a quest for innovation and collaboration. His story is one of embracing change in the pursuit of unlocking the secrets of how the human brain processes information, combining insights from different disciplines.

By Clelia Richters-Teodoro

Matthew's fascination with understanding the human mind began in childhood. "Since I was a child, I have always wanted to understand how I think, to understand how the thinking process works, how one thought leads to another. When I grew up, I was disappointed to find out that even the experts and the book authors didn't know the answer. So this is what I try to understand," he recalls. This early curiosity set the stage for a lifelong journey into the complexities of neural networks and artificial intelligence.

After a decade of working at a software company, he decided to return to academia, pursuing a PhD at Caltech in Southern California, to follow his dream of understanding the human mind.

His academic odyssey then led him to the Institute of Neuroinformatics in Zurich, a joint institute of the University of Zurich and ETH Zurich, where he led an interdisciplinary research group for 17 years, working in computational neuroscience, unconventional computing, and connectomics. Matthew's path then came to Groningen, drawn by CogniGron’s energised community and its supportive structures for encouraging interdisciplinary collaboration.

Transitioning to Groningen brought its own set of challenges, from the logistics of frequent travel to adapting to a new landscape. Yet, amidst the bustling streets and vibrant culture, Matthew found a sense of belonging and renewal.

Comparing Zurich and Groningen, Matthew is captivated by Groningen's relaxed and welcoming atmosphere. Conversations with fellow academics echoed his sentiments, affirming Groningen as a hub of creativity and contentment.

Looking to the future, Matthew envisions CogniGron as a nexus of collaboration, where diverse minds converge to unlock the deep potential of human-inspired information processing. He champions spontaneous interactions between researchers as the most important catalyst for innovation and progress, envisioning a future where academic boundaries dissolve and interdisciplinary visions bring groups together.

When asked about his contribution to CogniGron, Matthew explains: “Within the Cognitive Circuits and Systems theme, in our group we make conjectures about how the human cortex processes information. Based on these ideas, we build functioning systems that reproduce cortical capabilities such as visual processing, sensor fusion, abstract inference, and coordinated motor control. Through this process of conjecture and construction, we distill fundamental principles that enable neuromorphic algorithms to self-organise into systems exhibiting basic elements of intelligent behaviour.”

He further explains his dual approach: “On the one hand, I study how very simple components (less capable than a transistor) can form functioning systems with high-level computational behaviours. On the other hand, I study how neural networks can self-organise into systems exhibiting some sort of intelligent behaviour. As these neural systems show us what they can easily do, we learn to reimagine what computation is.” This dual focus underscores his commitment to merging the best aspects of engineered and naturally inspired systems.

Matthew also looks forward to working directly with the Materials Science groups in CogniGron. As he says, “The Cognigron environment is ideal for connecting us with researchers developing new materials, allowing us to analyse the computational potential of these materials.”

As we celebrate another year of advancement and discovery, Matthew's journey is a perfect representation of the unwavering spirit of CogniGron—a place where the future is unlocked, one breakthrough at a time.