I use Computational Psychiatry to understand schizophrenia and the psychosis spectrum. Computational models can link biological, social and psychological accounts of mental function and dysfunction in a mathematically rigorous way (Adams et al., 2016, JNNP).

There may be two major biological pathologies in schizophrenia: 

i) dysfunction of NMDA receptors, which may be compensated by interneuron downregulation, causing disinhibition and loss of 'SNR' in higher hierarchical areas such as prefrontal cortex and hippocampus.

ii) increased synthesis and release of dopamine in the striatum.

I use models of brain function to understand how these changes might contribute to schizophrenia and psychosis. Examples include:

i) Modelling perception, action & cognition: 

The brain may perform (or approximate) Bayesian inference on the causes of its sensory data (e.g. by updating top-down predictions with bottom-up prediction errors). Imprecise prior beliefs could cause various phenomena in schizophrenia (Adams et al., 2013, Front Psychiatry). Altered circuit properties could also affect belief stability and 'noise' in decision-making (e.g. in the ‘beads’ task; Adams et al., 2018, J Neurosci), and delusions could arise due to habitual and affective influences on a noisy cognitive system (Adams et al, 2021, Schiz Res). 

ii) Modelling imaging data:

Dynamic causal modelling (DCM) uses biophysical models of imaging data to estimate e.g. E/I in different brain areas. We found increased E/I in prefrontal cortex in schizophrenia (Ranlund, Adams, et al., 2016, Hum Brain Mapp) and decreased coupling between hippocampus and mPFC (Adams et al., 2020, Brain).

iii) Mapping behavioural models on to the brain:

Are our models of behaviour instantiated in the brain? We have tested whether behavioural model parameters correlate with E/I parameters from M/EEG data (Adams et al., 2016, Neuroimage) or dopamine 2/3 receptor availability from PET data (Adams et al., 2020, Cereb Cortex). 

iv) Discovering new psychosis phenotypes:

I have collaborated with Dr Alan Anticevic (Yale) in using DCM on large functional imaging datasets to try to understand excitatory and/or inhibitory neuron pathology in schizophrenia (Adams et al., 2022, Biol Psych), presented here: www.quentinhuys.com/tcpw/events/rick-adams/ 

My next project (2022-) aims to use both DCM and machine learning methods (applied to M/EEG data) to infer excitatory and inhibitory neuron pathology in individuals with schizophrenia and thus better target glutamatergic treatments early on in the disorder.

Google Scholar: https://bit.ly/33gpNdv

Semantic Scholar: https://bit.ly/2U1Q3nU