Connectome verification: inter-rater and connection reliability of tract-tracing-based intrinsic hypothalamic connectivity.

MOTIVATION: Structural connectomics supports understanding aspects of neuronal dynamics and brain functions. Conducting metastudies of tract-tracing publications is one option to generate connectome databases by collating neuronal connectivity data. Meanwhile, it is a common practice that the neuronal connections and their attributes of such retrospective data collations are extracted from tract-tracing publications manually by experts. As the description of tract-tracing results is often not clear-cut and the documentation of interregional connections is not standardized, the extraction of connectivity data from tract-tracing publications could be complex. This might entail that different experts interpret such non-standardized descriptions of neuronal connections from the same publication in variable ways. Hitherto, no investigation is available that determines the variability of extracted connectivity information from original tract-tracing publications. A relatively large variability of connectivity information could produce significant misconstructions of adjacency matrices with faults in network and graph analyzes. The objective of this study is to investigate the inter-rater and inter-observation variability of tract-tracing-based documentations of neuronal connections. To demonstrate the variability of neuronal connections, data of 16 publications which describe neuronal connections of subregions of the hypothalamus have been assessed by way of example. RESULTS: A workflow is proposed that allows detecting variability of connectivity at different steps of data processing in connectome metastudies. Variability between three blinded experts was found by comparing the connection information in a sample of 16 publications that describe tract-tracing-based neuronal connections in the hypothalamus. Furthermore, observation scores, matrix visualizations of discrepant connections and weight variations in adjacency matrices are analyzed. AVAILABILITY: The resulting data and software are available at http://neuroviisas.med.uni-rostock.de/neuroviisas.shtml.

Towards Differential Connectomics with NeuroVIISAS.

The comparison of connectomes is an essential step to identify changes in structural and functional neuronal networks. However, the connectomes themselves as well as the comparisons of connectomes could be manifold. In most applications, comparisons of connectomes are applied to specific sets of data. In many studies collections of scripts are applied optimized for certain species (non-generic approaches) or diseases (control versus disease group connectomes). These collections of scripts have a limited functionality which do not support functional and topographic mappings of connectomes (hemispherical asymmetries, peripheral nervous system). The platform-independent and generic neuroVIISAS framework is built to circumvent limitations that come with variants of nomenclatures, connectivity lists and connectional hierarchies as well as restrictions to structural connectome analyses. A new analytical module is introduced into the framework to compare different types of connectomes and different representations of the same connectome within a unique software environment. As an example a differential analysis of the partial connectome of the laboratory rat that is based on virus tract tracing with the same regions of non-virus tract tracing has been performed. A relatively large connectional coherence between the two different techniques was found. However, some detected connections are described by virus tract-tracing only.

["Who is the boss?"--The relation between social dynamics and learning in class]. / "Wer ist der Boss?"--Zum Zusammenhang zwischen der sozialen Dynamik im Klassenzimmer und Einstellungen zum Lernen in der Adoleszenz.

Research shows that antisocial behavior and learning are negatively related whereas prosocial behavior and learning are positively related, but evidence on how the social dynamics in class influence learning attitudes is non-existent. We were interested in tracking unsystematic differences in learning attitudes on a class level and how they relate to social impact based on dominance or social status. 1,159 pupils from 43 7th to 9th grade classrooms filled in a questionnaire on learning attitudes (TPB, Ajzen, 1991) and nominated their classmates on participant roles in bullying, resource control strategies, and social status. Based on hierarchical linear modeling we analyzed whether and how specific pupils influence the learning attitude of their classmates. Results show that the average learning attitude in class can be predicted by the most dominant individual. Nearly 9% of variance in individual learning attitude can be explained by group effects. The learning attitude of the individual identified highest on coercive and prosocial strategies and on social impact predicts 77% of the respective group variance. Educational implications need to focus on the psychological relevance of dominant children that may impede the developmentally appropriate progress of each individual in their classroom.