Profile

The main question of my work is how neurons manage to ‘talk to each other’ and communicate information – especially visual information. For example, imagine you are looking at a crowd, trying to spot the face of your friend. While you are doing that, lots of neurons in your visual cortex get activated, responding to bright and dark ‘pixels’ that they then glue together into different people’s faces, and other neurons are active signalling which faces are important to you and which ones you can ignore. So all these neurons are active together, and activate each other in different patterns, but we don’t know yet where in those activity patterns information is transmitted. It’s like hearing people talk without having any idea how human language works – clearly everyone is using their voice to make some kind of sounds, but what on earth are those sounds supposed to be doing?

People have been trying to solve this question for years by recording the activity patterns of neurons while a person (or an animal) is looking at things or solving a visual task. But there are so many interesting and possibly important neuron patterns – so to find out if an activity pattern really matters, we need to not only record it, we also have to change it and find out if the person or animal now sees things differently as a result. That’s what we do with our mice.

First we teach the mice to run towards ‘good’ visual targets in a computer game (usually stripey walls) instead of ‘bad’ visual targets (also stripey walls, where the stripes are at slightly different angles). Since the difference between good and bad targets is so small, the brain activity has to be pretty precise to be able tell the difference. We then record the activity of lots of neurons with electrodes that we stick into the brain, and for each moment we try to find the activity pattern that best predicts if the mouse is going to make the right or the wrong decision. The final bit is really cool and really new: a few years ago someone came up with a way of making neurons react to blue light. So whenever you shine blue light onto the brain those neurons become active. And with yellow light you can make them fall silent. This means that we can change the activity patterns of the neurons superprecisely – it’s almost like playing a ‘brain piano’. Depending on how and when we change the activity patterns, the mice get better or worse at the computer game, and that tells us which neuron activity patterns are most important in order to see the good and bad targets.

My days can be really different depending on what’s the most important aspect of work at the time. I usually get up pretty late because my brain doesn’t like early mornings. There are days when I do mostly experimental work – wiring electronics for new measurements, training mice on different tasks, doing surgeries to record neuronal activity from their visual cortex. On other days I sit somewhere (usually in a cafe close to campus) and write programs to analyze data and visualize them in different ways. We look at many different details of behaviour (e.g. how much the animals run in which direction, how often and how quickly they lick to get a soy milk reward, where they are looking…). All these details give us a clue of what the animal ‘might be thinking’, so it’s important to try lots of different analyses. On other days I mostly write up new results or plans for new experiments. I always work together with students, so I’ll interact with my students and try to teach them useful stuff throughout the day (though a lot of the time they end up teaching me instead 🙂 ). On most days there are one or two meetings to keep colleagues updated on each other’s work and/or on practical organization stuff.