Interviews with researchers

In the course Science Communication and Journalism the students conduct an interview with a researcher in the field of their topic. Two of the final interview products were published in the Behavioural and Cognitive Neurosciences magazine of November 2023.

An annual checkup for your brain, like that of a car, might sound like something from the future. However, if Hedderik van Rijn has something to say about it this might not be a dream for the future for much longer. His programme called Slim Stampen might be able to help with the early detection of memory problems. And earlier detection means a better chance at treating many of the diseases that display memory problems as a symptom.

Why an annual checkup?
Everyone who owns a car knows that your car must get checked out at least once a
year to keep it moving smoothly. In a complex machine like a car a lot can go wrong, and prevention is always better than a cure. Like a car our bodies and minds are also complex forms of machinery, however most people do not go to the doctor for an annual checkup and the ones that do only get a checkup on their bodies. While the health of the brain and mental health has gotten a lot more traction in recent years most of this attention has been on the treating of these problems, in particular diseases like Alzheimer’s. An early detection programme for Alzheimer’s disease would be a step in the right direction. But how would such a programme take shape and what is already being done now?

Memory Problems
The first signs of Alzheimer’s disease are small memory problems like forgetting your
keys when walking out the door, or forgetting to lock the door. While in and of itself
forgetting stuff like this might not immediately constitute a problem, it may be a cause of concern when coupled with other factors like age and genetics. When diseases like Alzheimer’s are discovered in patients there have already been many instances of the above-mentioned situations where routines or events are frequently forgotten. But even before this the so called “rate of forgetting” has already become higher. The “rate of forgetting”, or ROF, is the speed at which people forget information they have learned. While this rate is very personal there does seem to be an average for different age groups and people with memory problems deviate highly from these averages according to Hedderik van Rijn, Professor at the University of Groningen. According to van Rijn his program Slim Stampen (www.slimstampen.nl), which is normally used to help people study facts, could be used to measure this ROF and thus be used in the early detection of memory problems.

Why Slim Stampen?
“As with every child my age I was trying to understand my schoolwork with help from my mother” recalls van Rijn. “Most of the time she quizzed me during the washing and drying of the dishes when I was studying for a test”. The way that van Rijn’s mother quizzed him always stayed with him and was one of the cornerstones for the Slim Stampen programme. “I noticed that when I quickly answered question, for instance what is the capital of France, my mother would not ask me this question again” van Rijn continues. “However, when I had to think a bit before answering, even when it was correct, my mother would ask the question again later until I answered her as quickly as with the other items”. In his adult life van Rijn still used this way of quizzing with his own son and found studies that showed this way as very effective for learning facts. Prompted by these personal and professional insights van Rijn wanted to create a programme designed to do exactly what his mother did for him and help students learn facts in an optimal way. He wanted the facts to really stick, and thus created the programme called Slim Stampen.

How does Slim Stampen work?
“Slim Stampen utilizes a similar process as the one my mother used all those years ago” claims van Rijn. The process seems to be quite simple and on a very human level. To optimize the learning of the person that uses the programme it measures not only if a person answers correctly but also how long it takes them to remember and answer. “The longer someone takes to answer a question, the less frequent they actually know the right answer” van Rijn continues. “People often overestimate their own knowledge because of this”. An example of this is the question on capital cities. When you answer this slowly the person asking the question understands that you might not know this fact by heart, while you think that getting the question right means you know the answer even though you do not know it as well as you thought you did. Slim Stampen measures this rate and changes accordingly to make sure at the end you not just faintly know the facts, but you really know them by heart.

From learning to checkup
But how can a programme designed to help students learn facts be used as an early detection system for memory problems? Well, it turns out that it all comes down to the ability of Slim Stampen to measure the speed at which people answer. “When we let people use the programme with facts, they already know the rate at which they answer the questions shows us something about their rate of forgetting” tells van Rijn. “If this rate of forgetting falls between the normal parameters this is not a problem, however if the rate is quite high this might tell us something”. In other words, the way in which the Slim Stampen programme ensures that the person using it is learning as effectively as possible can similarly be used in the early detection of memory problems by using the rate of forgetting as a measuring stick.

The future
“Slim Stampen is not yet an early detection system for cognitive problems, and even though it is promising it will also probably never be used without other means and measures” says van Rijn. The programme should strictly be used to assist the current ways of diagnosing cognitive problems like Alzheimer’s and should always be used in a healthcare and professional capacity according to van Rijn. But before the program can be used in clinical practice more research is needed. “The next step would be to use Slim Stampen in a large population screening research programme to test its real-world ability to say something about cognitive problems” ends van Rijn. This means we still need to wait for some more time before a real annual health checkup for our brains becomes a reality.

Click to access bcn-magazine130.pdf

The release of ChatGPT was met with excitement, shock, confusion, and a good dose of uncertainty – especially in academia and education. Learning and language expert, Marije Michel, gives us her insights on how to navigate the murky waters of generative AI-powered language models. All while safeguarding what is most important -that students continue to learn.

A teacher walks amidst her students in the classroom. It is a German course for second language learners, and her students are busy with a written assignment. As she surveys her classroom, the teacher can see a particular software being used by all of her students as they work on the task: ChatGPT. Seeing this, the teacher feels no anger, no disappointment, no urge to step in. Why would she? She was the one who told them to use it.

The release of ChatGPT has been causing some concern to schools and universities. After all, the availability of a tool that can quickly produce answers and write small essays with apparent accuracy and appropriate language makes it hard for teachers to assess whether or not students have achieved their learning objectives. This, along with the potential for the software to be used for cheating on assignments or academic papers, has even led to ChatGPT being banned from some educational institutions.

But while ChatGPT and other AI applications can shake up the education system as we know it, this does not have to be a bad thing. That is the view of Marije Michel, Professor of Language Learning at the University of Groningen. “It’s a new challenge,” Michel told us. A challenge which presents new questions. “With ChatGPT being here, how can we improve learning? How can we ensure that the students still sit down and learn?”

Michel is currently researching how ChatGPT can be used to help students learn German as a second language. Alongside language teachers Iryna Menke-Bazhutkina and Niklas Abel, as well as researcher Carola Strobl, Michel asked students to read two German texts and summarise them in a third text, also in German. After the students were done, they could compare their work with that of ChatGPT, which had been asked to complete the exact same task. What the students saw was that despite the AI not making any grammatical mistakes, there were still some issues with its writing. “The overall coherence of a text written by ChatGPT was not so good,” said Michel, “mainly in the line of argumentation and in the structure
of the text as a whole.”

Michel’s expectation is that engaging with ChatGPT will help students to not only improve their language skills, but that they will also reflect on their experience. “We ask them what they have learned, what they can take from ChatGPT to, in the end, come up with a very good text. Where do you see that you, as a human, are actually better?”, Michel explains. “It is a great tool to train critical thinking.” At the time of writing, Michel and her collaborators have finished collecting data, and are now in the process of analysing it.

Despite the possibilities for new research and learning methods, Michel remains cautious about the negative effects that this technology can bring. Drawing from her previous studies on second language acquisition, Michel predicts that ChatGPT can cause some trouble, especially when it comes to learning how to be a good writer. “We know that good writers spend a lot of time with the planning and organisation of their texts,” said Michel, “and that is maybe why we see an outcry against ChatGPT.” She believes that coming up with your own thoughts and organising them is fundamental for learning how to write properly. “If you outsource that to ChatGPT, you won’t
learn from it.”

This learning process remains important, as there is still a place for good human writers in a post-ChatGPT world. The truth is, for Michel, cohesion and coherence are not the only aspects in which ChatGPT is lagging behind human writers. “It can be a marker of good language proficiency,” she explains, “that you know when to use elaborate phrases, when to use short phrases, when to repeat a word, when to use synonyms. There’s a lot of knowledge needed to do that well, and those are things that we typically don’t see that well being done in the generated texts”.

The reason why ChatGPT lacks the ability to “play around with the language”, the way humans do, is connected to the way in which the software functions. ChatGPT writes texts by predicting the string of words or sentences that are most likely to follow one another, in response to the prompt given by the user. For Michel, this process leaves out an important element of writing: the writer’s choices. “There are so many synonyms and different registers of language to use,” she says. “Deciding what to put where is not a matter of probability, it’s a matter of choice. And this choice is important as a matter of how you express yourself, your identity, but also how you keep the audience in mind.”

However, a proper understanding of writing and language is not only necessary when it comes to creating complex texts. Even when producing simpler texts with the aid of ChatGPT, it is important to have a good use of language. In this case, Michel predicts that ChatGPT may have a similar effect on writing as translation software has had on translating. “When Google Translate or DeepL came in,” she recalls, “the job of a translator changed. Now, a large part of it is what we call ‘post-editing’. They need to correct the translation.” And when it comes to this postediting process, Michel stresses that “you can only do that properly if you have enough knowledge.”

The problems with ChatGPT, however, are not limited to the ways in which it can interfere with learning and the consequences it may bring. It can also lead to unfair consequences when used irresponsibly. “It was unethical, in many ways, to release ChatGPT without preparing the world for it,” Michel warns. She explains that someone can use the software to pass a test and enjoy the advantages this brings, while others who have not used the software get left behind. “It can decide on people’s lives!”

ChatGPT may have brought many changes to the learning landscape, and many more are probably still to come. But for Michel, what matters most is to ensure that schools and universities remain places of learning, and she wil continue to do research and find new ways to improve learning. “I have a lot of ideas on how to use ChatGPT in class. I see tons of possibilities”, Michel delightedly explains, as she mentions the possibility of students using the chatbot as a sort of writing buddy who provides feedback. “That’s the privilege of a researcher: a new tool comes and we get to find out how we can make it work in education!”

In the end, for Michel, ChatGPT is “yet another tool we have to accept in our lives”. A tool which, like many others, requires a good deal of critical thinking and responsibility when being used. And being the tool that it is, Michel believes that it has the potential to bring about important benefits and make our lives a bit easier. For academics and students alike. “If students don’t have to be graded on papers because they were written on ChatGPT, maybe we can spend that time on oral exams, or debates. Maybe it frees more time to go back to human-to-human interactions,” she says, before concluding with a smile on her face, “and rather than grading those papers, I can just have a conversation again!”

Click to access bcn-magazine130.pdf

Leave a comment