CNS 2024 Q&A with Peter Kok
From daily illusions like seeing animal shapes in clouds or mistaking a curtain for a person in a dark bedroom to more complex ones, like the “hollow mask illusion,” (screenshot at right/above) our prior experiences and expectations shape how we perceive the world around us, sometimes in unexpected ways. Studying what happens in the brain during these illusions, cognitive neuroscientists like Peter Kok gain insight into the internal signals the human brain uses to shape reality.
“To properly understand perception we need to pay as much attention to internal signals, such as prior knowledge and expectations communicated by our memory systems, as we do to external signals coming from our eyes and ears,” says Kok of the Wellcome Centre for Human Neuroimaging at University College London. “Excitingly, we now have the tools to start to tease apart these signals.”
Kok is a co-recipient of the Young Investigator Award and will give an award lecture at CNS 2024 about his lab’s research using high-field fMRI and other technologies to understand how our expectations influence the way we see the world. I spoke with Kok about this work, new lines of research, and what he is looking forward to at the CNS annual meeting this April in Toronto.
CNS: What got you started in cognitive neuroscience?
Kok: I have always been fascinated and mystified by the question of how our rich conscious experience of the world arises from a lump of cells in our head. After doing both philosophy of mind and psychology courses, I decided that I had no idea how to answer this hard question, but that maybe by doing cognitive neuroscience experiments I could make some contribution to some of the “easy” problems regarding the relationship between the brain and the mind.
CNS: Why did you decide to research how expectations shape our reality?
Kok: After doing research internships during my Research Master at the University of Groningen vaguely on the topic of the neural mechanisms of subjective perception, I applied for a PhD advertised by Floris de Lange at the Donders Institute focused on the role of expectations in perception, and that’s where everything really clicked into place for me. I could not have wished for a more supportive supervisor and a more exciting project to work on. The reason I find expectations so interesting is that they give us a way to study situations where our subjective perception of the world differs from the signal relayed by our eyes.
CNS: What have been you and your team’s most significant contributions thus far to understanding how expectations shape perception?
Kok: Merely expecting to see a specific image can activate a neural representation of that image in the deep layers of the early visual cortex, even when nothing is actually presented to the eyes (Aitken et al 2020; Haarsma et al 2023). This is a remarkable phenomenon given that we used to think of the early visual cortex as a passive filter of signals coming from the eyes.
Of course just expecting to see something doesn’t cause you to see it (i.e., hallucinate). How are these signals kept apart? It’s very early days, but one of our recent studies suggests that these signals exist in separate cortical layers; while expectation signals are sent to the deep layers of the visual cortex, it’s the middle layers, which mainly receive feedforward signals, that correlate with what people subjectively perceive (Haarsma et al 2023).
Where do these expectation signals come from? Research from our lab and others suggests that a canonical memory structure, the hippocampus, is involved in both learning predictive associations (e.g. a specific sound predicts a specific image) and using these associations to activate representations of expected images (Kok & Turk-Browne 2018; Aitken & Kok 2022).
Apart from the scientific content, one of the things that makes me happiest as a relatively junior PI is when people in the lab spontaneously help each other with their experiments and analyses. Working with enthusiastic and kind people is such a treat.
CNS: What is the biggest question you still want to see answered in your field of study and how will new technologies help answer it?
Kok: It’s become evident that internal memory signals influence processing in the visual cortex. But how? How are these two different types of signals integrated, or weighted, and how does this result in subjective perception? This is a question that needs to be studied at a fine-grained spatial scale, using ultra high-resolution fMRI, and at a fine-grained temporal scale, combining machine learning techniques with EEG and MEG. This non-invasive neuroimaging in humans should be complemented with single cell recordings in animals, in order to really reveal the neural circuits underlying perception. Importantly, this requires using experimental paradigms that sensitively and specifically probe subjective perception. Even the most high quality neural measurements are useless unless they are accompanied by an experimental design that allows you to dissect the psychological phenomenon you’re interested in.
CNS: What’s next for your research?
Kok: So many ideas, so little time!
A lot of our current research focuses on whether the hippocampus, usually considered a pure memory region, actually influences the way we perceive the world. One avenue we’re taking is to study the communication between the hippocampus and the visual cortex by using ultra high-field fMRI to dissociate feedforward and feedback signals in the medial temporal lobe, which interfaces between the visual cortex and the hippocampus.
We have recently started investigating the temporal dynamics of expectation signals in more detail, using MEG, and we are planning to do more work to try to link neural oscillations to expectations and their effect on subjective perception. This also extends to the involvement of the hippocampus, which is a difficult region to measure with recordings at the scalp (e.g. EEG and MEG) but we are starting some collaborations that will hopefully allow us to test some of our questions in patients who have intracranial depth electrodes implanted for the purpose of epilepsy surgery. These intracranial recordings may also help us look for hippocampal signals in our MEG data, by giving us a better idea of what kind of signals to expect.
Finally, I am intrigued by the idea that perception itself has a memory component to it; after all, our neural signals always lag behind the reality that gave rise to them. This is most strikingly demonstrated by postdictive illusions, where a later input influences how an earlier one is perceived, such as in the cutaneous rabbit illusion. We are planning to investigate the neural correlates of postdiction using fMRI and MEG, which will hopefully shed more light on the way subjective perception is computed in the brain.
CNS: What are you most looking forward to at the CNS annual meeting in Toronto in April?
Kok: Having ping gai at the Queen Mother Cafe!
More seriously, I am looking forward to catching up with friends and colleagues from around the world, and hearing about the latest exciting research being done by the global cognitive neuroscience community.
-Lisa M.P. Munoz