Impaired cognitive abilities in spatial learning are among the earliest signs of Alzheimer’s disease and thus do not only provide insight into whether and how specific physiological and cognitive mechanisms are affected by the disease but also represent a way for identifying subjects potentially at risk of developing dementia in the future. Experimental manipulations providing important insights into how the disease affects specific brain structures and systems in most cases are limited to work on laboratory animals such as mice and rats. In turn, neurological and neuropsychological assessments in the clinic allow a detailed characterization whether and to what extent specific cognitive abilities appear to be changed compared to healthy controls. To bridge that gap, our aim is to bring classical behavioral testing paradigms such as the water maze into a clinical context by implementation as computer-based virtual learning tasks.
Behavioral science has to deal with a high variety of variables and complex relationships between them. Spatial learning in complex environments represents a significant challenge for the brain and involves a wide range of cognitive abilities. Modern statistical methods such as structural equation modeling allows estimating the influence of specific parameters and their relation to others in representing the functional processes that underlie the behavioral responses observed.
The group has three main scientific foci:
1. The Dresden Spatial Navigation Task as a sensitive tool to address impaired spatial cognition neurodegenerative diseases.
The Dresden Spatial Navigation Task (DSNT) is a computer-based virtual reality adaptation of the classical Morris water maze task. Our aim is to establish the DSNT as a flexible and easy-to-use tool allowing efficient back-to-back analyses of spatial learning processes in animal models and the clinic. This is based on using parameters that with regard to their conceptualization are nearly identical and address highly similar aspects of learning and memory in the physical water maze for animals as well as in its virtual counterpart for humans. Analysis of navigation performance is based on time-tagged xy-coordinates representing the moving path of a given subject for a respective trial. Aside the classical water maze measures (latency, path length, etc.) a number of new quantitative and qualitative parameters can be derived using Matlab, R, and Python-based scripting: number and position of repeatedly used landmarks, preferred heading directions, decision points, or main search strategy applied. Beyond greater effect sizes for classical parameters, the DSNT allows the identification of more specific learning deficits in a complex spatial learning challenge. In a first study we found a high sensitivity of the DSNT in discriminating MCI from early AD patients in terms of overall spatial learning performance as well as of performance in parameters specific to acquisition and integration of spatial information.
2. Role of region-specific neurotransmitter systems in mediating the reversal-phenotype after ablation of adult neurogenesis.
Learning the water maze task represents a complex challenge and the hippocampus is just a part of a network of structures such as ventral striatum, nucleus accumbens, thalamus, and orbitofrontal cortex that all together control the overall behavioral strategy, how relevant information is acquired into a representation, and how that representation is used for achieving the goal.
Measuring region-specific transmitter contents at different phases of the water maze task and combining the results with behavioral parameters into structural equation models dissect the dynamic and systematic relations in neurotransmitter (and metabolite) levels with the behavioral responses observed. Insight into how transmitter activity in defined brain structures corresponds to specific behavioral states and under different experimental conditions provides valuable information for understanding the roots of cerebral plasticity in the adult human brain.
3. Deciphering acquisition and expression of individual spatial knowledge using Markov chains.
Spatial learning essentially means to associate specific contexts with meaning that can be expressed as moving on, turn in a certain direction, or press a button. Since each human and each animal tested represents an individual agent, we observe highly individual movement trajectories, providing (or preventing) specific experiences to be made, thus forming an individual learning biography. While for some subjects tested a group of landmarks is recognized as salient they will gain no or only low relevance for other subjects. In the first case, the landmarks will likely be part of a representation while in the latter landmarks might be encoded –if at all– only implicitly.Quantifying how differentiated a subject interacts with spatially specific contexts in a previously unknown (and meaningless) environment during and after a given number of acquisition trials provides a sensitive and specific measure of how spatial knowledge is acquired and used in different contexts.
In this project we use a high-dimensional Markov chain approach to describe an individual’s cognitive map used for complex behavior such as spatial navigation. This allows a non-invasive but still extremely specific assessment of how spatial information is acquired, integrated, and used. A particular strength of the described method lies in the fact that it yields the same kind of output metric for human and animal subjects, further strengthening our translational approach.