Matt Lambon Ralph is the Professor of Cognitive Brain Sciences. He is also the Unit Director of MRC Cognition and Brain Sciences Unit at the University of Cambridge. You can read his detailed biography and publications here. His research interests include semantic cognition and its disorders, aphasia, recovery, rehabilitation and neuroplasticity.
This interview was given by Dr Matt Davis on behalf of Prof Matt Lambon Ralph. Dr Davis has worked with and personally known Prof Lambon Ralph (LR) for a long time. Dr Davis leads 'Adaptive processing of spoken language' programme at the CBU. You can find more about Matt here. Matt was in conversation with Shrankhla Pandey, a PhD student at the University of Cambridge.
Could you share the key questions or challenges his work seeks to address?
So, the first big question Matt’s (LR) work looks at is semantic cognition, basically, how we know what we know about the world. How do we know the names, uses, and functions of all the people, objects, and places we come across in daily life? The second major area is about understanding language impairments, what happens when someone has a stroke, a brain injury, or develops dementia. He tries to figure out the nature of the cognitive and neural problems they face, and what those issues can tell us about how the brain works when it’s healthy. Then there’s a third part, which is about using what we’ve learned to actually help people. So, we take that understanding of the brain and mind and try to support recovery and rehabilitation, especially for people with aphasia, brain injuries, or dementia. The idea is to help improve their language and semantic cognition.
And when you put those three together, they really do fit nicely. Matt tends to start with looking at how that goes wrong in conditions like aphasia or brain injury. Then he returns to the basic science, trying to understand a key part of language or meaning. And then he goes back to the clinic, where you try to figure out why some people recover and others don’t, and do what you can to help improve outcomes.
It’s kind of like a three-legged stool, each leg supports the others, and together they make stable base. That way, you’re not just doing research for the sake of it but actually helping people with language impairments while learning more about the brain and mind.
Matt’s work has always had a close connection with clinicians, neurologists, neuropsychologists, speech and language therapists, and others, who work with people dealing with language problems. His own PhD was in neuropsychology and since then he has also become an honorary speech and language therapist (by the Royal College of Speech and Language Therapists). And thus, those patients aren’t just subjects of study. They’re actually a big priority for Matt when he thinks about the research he wants to do. He’s curious about the questions that matter the most to them.
He also thinks about what kind of knowledge would be most useful, not just for the patients, but for the clinicians who support them. That way, the research can really make a difference in the long run. It’s a great example of translational research, the kind that doesn’t just stay in the lab or in academic journals, but actually gets turned into real-world benefits for individuals, their families, and carers.
What shaped his journey into language science?
Prof. Matt Lambon Ralph actually had other careers before diving into science and research. He worked at Coutts & Co., dressed in frock coat, really incredibly smart. So, from working in the most upper-crust financial institution in the UK, he made a big shift and went to university to study psychology. And then he got bitten by the curiosity that comes from encountering patients with language impairment. That was a turning point.
Matt (LR) and I (Matt Davis) work at opposite ends of the lifespan, he focuses on language in later life, and I work on language development in early life. But what brought us together was a shared inspiration from a type of computational theory that was popular in neuroscience and psychology in the late 1990s: neural networks.
Neural networks are everywhere today, they’re the talk of AI, but the theory actually comes from psychology. For example, Geoffrey Hinton studied natural sciences, did his PhD in psychology at Cambridge and worked at the then Applied Psychology Unit (now the CBU) in the early 1980s; Hinton even received a Nobel Prize last year for his work.
Before neural networks, the field leaned heavily on symbolic AI and memory systems in the 1970s. But those had limitations, especially in handling variability and learning from experience, which neural networks were better at.
Take semantic dementia, for example, a neurodegenerative condition where people struggle to remember rare or less common words. That pattern fits naturally with how neural networks behave: weaker patterns get disrupted, while stronger ones remain. It’s a great example of how theory and real-world conditions align.
Over the years, Matt has worked with some truly inspiring people. Two names come to mind: Jay McClelland and Karalyn Patterson.
Jay McClelland was one of the key figures in the 1980s who helped promote neural networks, not just for psychology and neuroscience, but also for building AI systems that solve real problems. Based at Stanford and with DeepMind in London, he is still highly active in the use and the pioneering of foundational machine learning today. Matt first learned and started to use computational models in his final undergraduate years and in his PhD. He was immediately interested in the mechanistic formulism that they provide, including for thinking about patients’ language disorders. Matt and Jay have published together and still collaborate. They first started to collaborate whilst Jay was still at Carnegie Mellon in Pittsburgh, and Matt came to Cambridge to work with him and Karalyn Patterson in order to explore semantic disorders and use the models to simulate the results.
Karalyn Patterson is probably the foremost neuropsychologist in the UK, maybe even globally, especially when it comes to understanding how brain injury and neurodegeneration affect language. Matt came to Cambridge in 1997 as a postdoc to work with her as she had a large NIMH grant with Jay, and she’s been one of his most important mentors ever since. Karalyn’s retired twice now, but she’s still a huge figure in the field. After her first retirement from the CBU, due to the then mandatory MRC retirement age, she volunteered to work as a neuropsychologist in the specialist Early Dementia Clinic at Addenbrooke’s Hospital, helping patients with memory and language impairments. She did that for over a decade before deciding to slow down a bit. She’s still active, attending talks, publishing papers, and contributing to research, even with the most advanced brain imaging methods. What’s amazing is how adventurous she’s been with new techniques. Her background is in psychology and behavioural testing, but she embraced everything from computational modelling with neural networks to PET, MRI brain imaging, and electrical activity measurements. Every time a new method came into the field, she found ways to use it and collaborate.
And I think that spirit of curiosity and openness to new technologies is something Matt shares. He gets genuinely excited about how new tools and methods can push our understanding forward and ultimately help patients who are struggling.
What does his day-to-day look like?
Even though Matt (LR) has a packed diary, he still makes it a priority to carve out time for small group and regular one-to-one meetings with the researchers who are doing the hands-on work. That’s really important to him. He’s driven by a desire to stay close to the research, to tackle challenges alongside students and researchers, and to celebrate the victories when they come. As Director of the Unit, he’s responsible for overseeing a lot of interesting and important work. But what stands out is how hands-on he is in guiding and supervising his team. Even with junior PhD students, he’s generous with his time.
He is the director of an institute with huge demands on his time. But actually, the research is the foundation and the motivation. That's what gets him out of bed in the morning, even on a Monday.
What societal changes he envisions his research contributing to?
Aging and dementia are huge societal challenges, not just for the UK, but for all modern societies. We’re living longer than we did 50 or 100 years ago, and the challenge now is making sure we live those extra years in good health. We understand a lot about how to look after physical health, but cognitive and neural health are a vital challenge. It is challenged by people who survive strokes or dementias, but they’re often left with disabilities that deeply affect their day-to-day lives; whether it’s struggling to care for themselves or losing the ability to have meaningful conversations with loved ones.
There’s definitely space for medical understanding and technological support to play a bigger role here. For example, we still don’t have a cure for the types of dementias that affect speech and language most severely. There are some candidate drugs, especially for Alzheimer’s, which show promise in tackling the early stages and slowing down decline. But that also means we need to detect these conditions much earlier, before the damage builds up. And that’s not just a job for medics and biochemists, it’s also a challenge for psychologists and neuropsychologists, who are working to spot the subtle signs that someone’s language or memory is starting to fail.
If we can catch those signs early, we can intervene while there’s still time. That’s one way research can make a real difference.
Another example is in stroke care. Medical treatment for stroke has improved a lot in the last couple of decades, people survive strokes that would’ve been fatal before. But that also means more people are living with serious disabilities, including problems with movement, language, and semantic memory. So now there’s pressure to develop new forms of rehabilitation to support them.
And here’s where technology might really help. Think about how long it takes to learn language, years and years. If someone has a stroke that severely disrupts their language, they might need just as much time and support to recover from it. And increasingly, we can outsource some of that support to technology, like AI systems that automate spoken language interactions. It’s an exciting possibility that these systems could play a real role in rehabilitation.
Do brain implants face greater resistance and lower societal acceptability compared to other implants like pacemakers?
That’s a really interesting question, and it can lead to philosophical territory pretty quickly. One thing I find fascinating is how we still talk about the heart when we talk about love and romance, or the stomach when we talk about fear and anxiety, like getting butterflies when we’re nervous. We tend to place emotions and sensations in different parts of the body, even though we know that these experiences are actually functions of the brain.
So, if someone has a pacemaker, we don’t assume it affects their emotions or their ability to feel love. Of course not. But when it comes to the brain, there’s still a lot of resistance, because the brain feels like it’s who we are. It’s the story of our experiences, our personhood. There’s even a joke that a brain transplant is the one surgery where it’s better to be the donor than the recipient, which says a lot.
So, when we talk about neural implants, it raises important questions. Are we supporting someone’s brain function, or are we controlling it? That’s a challenge for the people developing these technologies, to show that they’re enabling disabled individuals to do what they want to do, not turning them into puppets of someone else’s intentions.
And I think this isn’t a new challenge for neurotechnology. Let me give you an example, not from my work or Matt’s, but from our colleague Alex Woolgar at the CBU. Alex works with individuals on the autism spectrum, particularly those with more severe forms who are non-verbal. There’s been a long and controversial debate around “facilitated communication”, the idea that these individuals might have movement disabilities as well, and that with the right support, they could use a letter board to point, spell, and communicate.
But the controversy is whether it’s really the autistic person communicating, or the facilitator. It’s a bit like a Ouija board, where the pointer moves, but no one claims responsibility. So, there’s been suspicion: is the communication genuine, or is it being guided by someone else?
And I think that’s exactly the same challenge we’ll face with neural technology. We need to make sure its giving voice to the person, not putting words in their mouth. That’s a really important challenge. I don’t know the perfect solution, but I do think the history of working with disabled individuals gives us a foundation. The right approach is to be cautious, to acknowledge limitations, and to be open to discussion.
That’s also part of what Alex is doing, recording brain activity to understand how much speech and language comprehension these individuals have, and running stringent tests to ensure that when communication happens, it’s truly coming from the person.
What is his ambition for the field of language sciences as we look toward 2050?
Matt’s been doing some fascinating research with neurosurgeons in Japan, where they use a technique involving temporarily implanted grid electrodes in the brains of epilepsy patients. These electrodes allow for incredibly detailed monitoring of electrical activity, which helps guide neurosurgery. For patients who are candidates for brain surgery, this is a vital part of the process. It helps surgeons understand which areas of the brain are still functioning well, and which areas are causing seizures, so they can remove the problematic tissue while preserving language and other cognitive abilities.
What’s really exciting is that this is cutting-edge, state-of-the-art science. But at the same time, there are companies, especially in Silicon Valley, who are looking at these technologies not just for temporary use in surgery, but as long-term solutions for neurological conditions.
Take Parkinson’s disease, for example. One treatment involves neural implants in the basal ganglia, where neuron degeneration affects movement. Electrical stimulation in that area can restore fluent movement to someone who might otherwise suffer from severe tremors or even be frozen, unable to move. It’s an amazing possibility, and it opens the door to similar neurotechnology that could help support or rehabilitate speech and language function. Imagine giving voice back to someone with a severe language disability. We’re not there yet, but it’s not hard to imagine that we could be in 25 years.
I have a friend at Stanford working with patients who have ALS, what Americans call motor neurone disease. Think of Stephen Hawking, who had a progressive disability but could still communicate through very small movements in his cheek, or breath. But there are patients with ALS who can’t move at all, not even voluntarily, and can’t communicate.
Now, some of those patients are being fitted with implanted electrodes that record brain activity and connect to computers that convert that activity into speech. It’s incredible, giving voice to someone who otherwise couldn’t speak at all.
So yes, looking toward 2050, it’s exciting to imagine that these kinds of technologies could become routine, not just experimental, but part of everyday clinical care.