Computational psychiatry is a rapidly growing field at the intersection of many fields, involving people scattered across continents.

The Transcontinental Computational Psychiatry Workgroup is a monthly web-based meeting in which we hope to foster discussion and exchange between those involved. We also organize a yearly satellite meeting at the Society of Biological Psychiatry. Here is a picture of the first meeting in San Diego in 2017:

The specific aims for the work group are:

To identify important problems in psychiatry, which are amenable to a computational approach.

This aim will determine “low hanging fruit”, i.e. challenges which have plagued psychiatric research and which may now be amenable to quantitative treatment (e.g. reward circuitry and addiction, fear circuitry and anxiety, cognitive control and schizophrenia).

This aim will help to set up programs of research that will subsequently seek out partnerships with clinical institutions to be able to quickly implement studies.

To develop collaborations across laboratories to be able to implement approaches quickly and to be able to address particular questions in a timely manner

The ultimate goal of this aim is to develop consortia that are focused on particular questions, disease, or processes that are of high relevance to psychiatry. For example, it is very clear that psychiatric disorders are highly heterogeneous, individuals labeled as having anxiety or depression might actually belong to many different categories or exhibit a variety of different dysfunctional processes. Similar to genetics, now imaging, and most likely behavior, heterogeneity implies the need for large patient samples. This is not possible within one group. Therefore, we will have to work together to implement similar paradigmatic approaches across laboratories.

To discuss various computational approaches, their implementation, their advantages and disadvantageous and determine how to improve these approaches for applications in psychiatry.

It is likely that modeling individual’s behavior will require a number of different computational approaches. Moreover, to appreciate individual differences we will need to obtain robust parameter estimates for individual subjects for different processes. Several computational approaches have already been developed for this purpose. It will be necessary to compare and develop these approaches across laboratories.

Feel free to add your thoughts as comments below, to write to us or indeed to sign up.

Michael Browning, Quentin Huys and Martin Paulus

Leave a Reply