"This is us" - Questions for Anika Löwe

December 06, 2023

Our institute has over 300 employees. But that is just a number. Who are the people at our institute? What do they do and what drives them? In our "This is us" format, colleagues answer questions about their work and their motivation.

On the occasion of International Women's Day on March 8, 2023, we started the series “This is us” with 15 female scientists from our institute. We are picking up where we left off and introducing the scientist Anika Löwe from the MPRG NeuroCode. In the last episode we introduced Sonja Sudimac from the Lise Meitner Group for Environmental Neurocience.

In the MPRG NeuroCode, you deal with so-called "aha moments" and the factors that trigger them, among other things. What fascinates you about this topic?   

What I find particularly fascinating about "aha moments" is that we are usually unable to trace back the cognitive processes that lead to them - the new solution or connection seems to jump into our consciousness suddenly, as if by magic. I am also fascinated by the question of what factors cause our brain to instantly start absorbing or restructuring information that has often been present for a long time but has been blocked out by attention processes. Which factors lead to our focus being interrupted so that an "aha moment" can occur is one of the main interests of my research.  

To what extent can these "aha moments" also be evoked in machines? What does this mean in relation to humans?  

We have trained neural networks to perform the same decision task that we use to explore "aha moments" in humans. In this task, there is a hidden strategy that allows the task to be performed much more efficiently and easily when humans have an "aha moment" about this strategy. Machines, of course, do not have a conscious "aha moments", but exhibit all the behavioral characteristics that we observe in humans: the switches to the hidden strategy happen very suddenly, only a proportion notices the hidden strategy (about 50% in both humans and networks), and the time at which the aha moment occurs varies equally between humans and networks. This implies that "aha moments" may not be such a unique, "magical" phenomenon, but a special kind of learning that can occur even in very simple systems without specific awareness of it.  

Furthermore, in the machines, we can examine the parameters of networks that exhibit aha-moment-like behavior in more detail and draw conclusions about what led to them. We found that a combination of three factors leads to aha-moments in networks: 1) "dormant knowledge" about the hidden strategy before it becomes relevant, 2) neural noise that we add to each update of the network, and 3) the regularization of certain network weights, which simulates an attentional process. This finding in turn generates interesting hypotheses for possible mechanisms of "aha moments" in humans.  

What influence could sleep have on these "aha moments"?  

Sleep plays a potentially interesting role in "aha moments", as information restructuring and memory consolidation take place during sleep. The first sleep phase of the cycle, N1 sleep, could be particularly important here, as the hypnagogic experiences during this sleep phase are considered an enhanced version of mind wandering. The physiological characteristics of N1 sleep could lead to new ideas being generated while logical inference skills are still being preserved.  

I am currently investigating this in collaboration with Marit Petzka, postdoc in our group, by using the same "aha moment" task described above, but having subjects take a nap halfway through the task. We determine the sleep phases using EEG measurements and investigate whether aha moments occur more frequently after (N1) sleep.  

Where do you get your research ideas from?  

A few years ago, my supervisor, Nico Schuck, reported interesting MRI results on "aha moments". These inspired me to investigate the factors behind "aha moments" in more detail. The question of whether a gradual learning system, which is a neural network, can show aha-moment-like behavior arose from conversations with researchers at Oxford University, which led to a wonderful collaboration and a research stay for me at the Gatsby Center at UCL. In general, what I find incredibly exciting and motivating about "aha moments" is that everyone knows and has experienced them, but so little is known about how they occur.  

What challenges do you face as a scientist?  

It's no secret that female scientists are still very disadvantaged today and often must assert themselves much more strongly and in a completely different way than their male colleagues. Especially in the more mathematical parts of neuroscience, where I work, the ratio is changing drastically. Especially after the PhD, many female researchers leave academia, which is partly due to the difficult conditions for starting a family, from which female researchers suffer much more. However, there are many opportunities for improvement for female researchers that we should all work towards together.  

What do you appreciate about the Max Planck Community? 

I consider myself very lucky to be a PhD student at a Max Planck Institute. The exchange between different groups and institutes, but also the international connections are special and very beneficial for joint scientific progress. In addition, I have always received great support and all the resources for my research, which I have been able to design and carry out entirely according to my individual wishes. You feel part of a large network of incredibly inspiring people. 

Go to Editor View