Joseph Balfe
Joseph Balfe
BMus, BSc, MSc (Distinction)
DPhil (PhD) Student
- Clarendon Scholar 2025
- Rachel Conrad Doctoral Scholar 2025
Toward understanding depressive cognition using non-invasive brain stimulation, multimodal neuroimaging, neuropsychological testing, and computational modelling
Research
Background
The brain elegantly orchestrates an interplay between top-down predictions and bottom-up prediction errors to guide effective perception and action in the face of uncertainty. The fine balance between these processes yields an approximation of the world that is continuously updated in a Bayesian manner. The precision-weighting, or "tuning," of such predictions and prediction errors determines their influence and behavioural relevance. In other words, do I trust my prior beliefs, or my new information - and by how much?
Advancements in computational neuroscience suggest that psychiatric conditions can be accounted for by aberrant predictive processing machinery, whether it be top-down predictions, bottom-up prediction errors, or the precision-weighting of either. In depression, highly precise negative top-down predictions are thought to maintain persistent negative affect, rendering new bottom-up information as largely irrelevant for model revision. In this context, depression can be considered as a disorder of maladaptive belief updating.
PhD Research
Given the neurobiological nature of depression, it is no surprise that successful antidepressant treatments have one thing in common: neuroplastic induction. By inducing a heightened state of plasticity, a ‘window of opportunity’ presents itself to enable the reorganisation of how information is processed within the brain’s hierarchical architecture in real-time. As such, non-invasive brain stimulation methods like transcranial direct current stimulation (tDCS) and transcranial ultrasound stimulation (TUS) hold promise for the targeted modulation of such information processing.
My work focuses on the real-time modulation of two separable but interacting inferential processes: (i) the precision-weighting of prediction error signals, which determines their influence on belief updating, and (ii) the credit assignment of these precision-weighted prediction errors. By using the above brain stimulation methods, I hope to better understand how we can normalise aberrant predictive processing machinery toward more adaptive and accurate modelling of the world.