Cognitive and Neural Mechanisms of Change
Neural and Cognitive Mechanisms of Change Across Multiple Timescales
Our daily lives are shaped by a variety of learning and adaptation processes, many of which unfold over weeks or even years. Children, for example, typically need 2–4 weeks to learn how to ride a bike, amateur chess players require months to significantly improve their skills, and individuals undergoing psychotherapy sometimes need years to redefine their behavior and experiences.
These long-term learning processes raise important questions: How do shorter episodes—such as a single chess lesson—contribute to the long-term development of strategies and adaptation processes? Psychological and neuroscientific research has shown that, in many cases, humans are particularly skilled at generalizing from individual experiences to draw lasting conclusions. In fact, we possess the ability for "meta-learning"—the capacity to learn how to learn more efficiently.
Studying Learning and Change Across Time
The research focus on Neural and Cognitive Mechanisms of Change investigates learning and adaptation processes over both short and long timescales using a combination of approaches from psychology, neuroscience, and artificial intelligence research. We study factors that modulate learning efficiency and other aspects of learning, such as how early developmental learning influences long-term outcomes or the neural mechanisms underlying psychotherapy.
Neuroplasticity in Sensitive Phases and Development
A central theme of our research is neuroplasticity and its modulation. Early developmental periods are characterized by "sensitive phases"—windows of time when the brain exhibits heightened plasticity and is particularly receptive to experiences. Adequate experiences during these phases are crucial for typical neurocognitive development. While sensitive phases for emotional and social processes are less well understood, emerging studies suggest their existence. For example, early childhood trauma is more strongly linked to later psychopathology than trauma occurring later in life. However, the specific processes that may be subject to sensitive phases in humans remain unclear.
MoC research investigates topics such as how sensory deficits in early childhood affect development, the role of pre-verbal statistical learning in shaping children's social behavior or why many mental health disorders emerge predominantly during adolescence.
Stress, Sleep, and Reactivation as Modulators of Plasticity in Adulthood
Recent research suggests that even in adulthood, there may be other factors that induce brief time windows in which certain experiences can induce heightened neuroplasticity. The reactivation of conditioned fear can for instance create a temporary window for permanently erasing fear memories, and physical exercise has been shown to briefly enhance neuroplasticity. But also exposure to stress may induce a time-limited period of enhanced memory formation for ongoing events, and sleep also plays a crucial role in long-term memory consolidation, as demonstrated by numerous studies. Several MoC projects investigate how specific experiences, reactivation of memories, stress, and sleep affect short- and long-term learning processes.
What Can We Learn from Artificial Intelligence Research?
In recent years, artificial intelligence research has made remarkable progress, offering key insights into learning mechanisms at an algorithmic level. How do models like ChatGPT learn, and what similarities and differences exist between artificial and human learning? Are there "mathematical laws of learning"?
To answer these questions, some MoC projects integrate machine learning models, such as neural networks, to better understand how learning might be implemented in complex systems like the human brain.
However, current AI models struggle with the interplay between short- and long-term learning processes. For instance, neural networks often exhibit catastrophic forgetting—they tend to erase old knowledge when encoding new information. In contrast, humans can continuously build upon past experiences and derive lasting, generalizable conclusions. Understanding these parallels and contrasts between human cognition and mathematical learning models is an increasing focus of MoC research.