Prof. Dr. Gesine Dreisbach completed her intermediate diploma in psychology at the University of Mannheim in 1992 and her diploma in psychology in 1997 at the Technical University of Berlin. In 2000 she received her PhD in Dr. phil. University of the Federal Armed Forces, in Hambur.
From 1997-2000, Ms. Dreisbach was responsible for research in the area of “cognitive activity” at the University of Federal Armed Forces, Hamburg. This was followed, among other things, by research at the Technical University of Dresden at the Chair of Psychology and a professorship at the Johann Wolfgang Goethe University in Frankfurt / M. Since 2009 she has been a professor at the Chair for General Psychology at the University of Regensburg.
In her research, Ms. Dreisbach focuses on processes of cognitive control. Cognitive control enables people to adapt thinking and acting dynamically to changing goals and task rewuirements.
In addition, Ms. Dreisbach takes on editorial work in the field of experimental psychology and is a series author for SpringerBriefs in Cognition. As an eypert for research institutions, she works for the German Research Foundation (DFG) and the German Academic Exchange Service (DAAD), among others. She is also a reviewer for scientific journals such as Acta Psychologica, Journal of Experimental Child Psychology, Journal of the International Neuropsychological Society, The American Journal of Psychology and many more.
Chair for General and Applied Psychology
Chair of General and Applied Psychology, Faculty of Human Sciences.
Learning, motivation, emotion (basics BSc), human-machine interaction (application BSc), cognitive control and executive functions (master).
Computaional models of human processing (master seminar on executive functions), models of adaptive action control (keywords: context sensitive control, conflict monitoring).
The focus of my teaching is not computaiona / mathematical modeling but the wuestion of which cognitive mechanisms are based on these model assumptions and which experimental paradigms of cognitive psychology are suitable for empirically checking the model assumptions.
Theoretical and practical foundations of experimental cognitive psychology.
I think it is important to make it clear that the foal of cognitive psychology is to understand the cognitive (and neurophysiological) mechanisms of human action control. In AI, it seems to me more about the funcionality of the result. The two disciplines can methodically benefit from each other.
In my research, I deal with questions of adaptive action control. In a dynamic environment, we are constantly faced with antagonistic requirements. On the one hand, we have to be able to concentrate on a current task and to shield it from disturbances. On the other hand, we also have to react flexibly to environmental changes. My research focuses on the question of how this balance between cognitive flexibility on the one hand and cognitive stability on the other hand is modulated.
In the broadest sense, conflict monitoring theory (Botvinick et al., 2001, Conflict monitoring and cognitive control) is most closely related to AI. Much of my research is based on this computational model. The model tries to explain how a conflict of action is resolved. A conflict of action would be e.g. for example: I see the word GREEN and am supposed to name the color in which the word is written. A conflict of action exists because the automated and therefore dominant process of reading (green) has to be suppressed in order to name the color (red) in favor of the actual goal. The success of the Conflict Monitoring Model (as of today, 6841 citations, first author Matthew Botvinick is now director of Neuroscience Research at DeepMind, London, UK) is due ti the fact that cognitive control, which is necessary to solve this conflict of action, does not have a central one “Decision maker” is exercised. Instead, it is assumed in the model that the activation of two mutually exclusive actions (in the example: pronouncing “red” vs. “green”) is measured as energy above these output layers. This conflict via the output nodes is detected (by a conflict detector), which then sends signals to a task demand unit, which in turn strengthens the task representation by the input layers in the subsequent passage (new color word) being more sensitive to color than word information. This means that behavior control is triggered by the detection of a conflict, but this detector itself does not need to know what is right or wrong. This computational model is supported by behavioral data in such a way that it can be shown that test subjects react fater to a second conflict stimulus (solve the conflict faster) if they are preceded by a conflict than if they were not preceded by a conflict(e.g. if the previous stimulus was GREEN). And at the neuropsychological leve, there is evidence that the conflict detector is located in the anterior cingulate cortex and the task demand unit in the dorsolateral prefrontal cortex. The example shows how computational models, imaging and experimental behavioral research complement each other.
I have no practical knowledge of AI, i.e. I have no modeling experience. But I use these models to predict (and sometimes explain) behavior. I was a postdoc for one year with Jonathan D. cohen in Princeton, who combines these three pillars (computational modeling, imaging, experimental behavioral research) in his lab.
That depends on how far / narrown you can get AI. To be honest, it’s not that clear to me here.
As soon as I see parallels to my own research topics (adaptive cognition), I’m interested. And I think that all artificial systems have to be adaptive if they are to be successful (or otherwise end up on a wall like the Duracell male from advertising). And that’s exactly the part that interests me in human behavior. The incredible flexibility that we display without being completely chaotic.
I carried out the last project with a young scientist from biology. His speciality is the behavior of ants. We have investigated and compared how conflicts between humans and ants are resolved. The surprising finding that ants can also resolve conflicts does not show the sequential flexible adaption to conflicts described above from humans.
Since AI can be used practically (again: I’m not sure how narrowly / broadly you use the term here) everywhere, I think grater awareness is important. It is difficult for me to make another forecast. However, I believe that AI can benefit greatly from being close to psychology. In other words: It will not be possibly to artificially create intelligent behavior with a mere increase in computing power (it is clear to me that today’s AI no longer see it that way).
Honestly, this was the first time I was on the site, and was surprised to find my name there (because I see my connection to AI, as it should have become clear, rather indirectly). Now that I have answered the questions, I have the impression that I could fit in quite well with my research focus.
A somewhat clearer definition of what is meant by AI here would be helpful.
Ms. Dreisbach, thank you for taking the time to answer these questions. We wish you a nice day!