Full Program »
Hallucinations Result from the Overweighting of Perceptual Priors
Background: Some people hear voices that others do not, but only some of those people seek treatment. Those that need treatment are more distressed by their experiences, they suffer more and are at risk of harming themselves or others. This risk is particularly acute for the 10-30% of patients whose hallucinations are intractable to current treatments. There is a clear need for a deeper more mechanistic understanding of hallucinations, one that links the biology with the phenomenology of voice hearing. By conceiving of the mind and brain as inference machines, bringing past experiences to bear on current sensations, we can perhaps understand hallucinations in terms of an imbalance in this process such that prior beliefs are over-weighted and percepts are created where there should be none. We sought to test this hypothesis.
Methods: Using a Pavlovian learning task during functional neuroimaging, we induced conditioned hallucinations – driven by learned expectations in four groups of people who differed orthogonally in their voice-hearing and treatment-seeking statuses. We captured behavioral and brain responses to this training and analyzed those responses using a hierarchical computational model whose parameters we estimated on a subject by subject basis using Bayesian model inversion. In separate studies, we manipulated the regions identified using transcranial magnetic stimulation.
Results: People who hear voices were significantly more susceptible to the effect. Using functional neuroimaging and computational modeling of perception, we identified processes that differentiated voice-hearers from non-voice-hearers and treatment-seekers from non-treatment-seekers and characterized a brain circuit, incorporating the anterior insula, superior temporal sulcus (STS) and auditory cortex that mediated the conditioned hallucinations. Critically, voice-hearing participants over-weighted their priors and this was associated with stronger insula and STS responses. Furthermore, people with a diagnosed psychotic illness did not develop beliefs about task volatility and therefore did not update their priors. Manipulating STS with inhibitory 1Hz TMS significantly attenuated voice hearing (and global symptom scores) in individuals with intractable voices. Furthermore, connectivity between STS and the insula improved as a function of symptom improvement
Discussion: These data demonstrate the profound and sometimes pathological impact of top-down cognitive processes on perception and may represent an objective means to discern people with a need for treatment from those without. In addition, these computational approaches have inspired new candidate treatments for hallucinations as well as behavioral and neuroimaging means to match individuals to those treatments.