Perceptual expectations and false percepts generate stimulus-specific activity in distinct layers of the early visual cortex

Abstract

Popular theories suggest that hallucinations arise through excessive top-down perceptual expectations, whereas others have emphasised the role of spontaneous bottom-up activity. These theories make different predictions about how input and feedback layers in sensory regions contribute to hallucinations. Here, we used layer-specific fMRI to interrogate neural activity underlying hallucinations - high confidence false percepts - and perceptual expectations while healthy participants (N=25) performed a perceptual discrimination task. We found that false percepts were related to stimulus-like activity in the middle input layers of V2. On the other hand, perceptual expectations activated the deep feedback layers of V2, without influencing perception. The prevalence of high confidence false percepts was related to everyday hallucination severity, confirming their ecological validity. These results reveal that hallucinations can arise through spontaneous stimulus-specific activity in the input layers of the visual cortex, independent of top-down perceptual expectations.

Expectations about presence enhance the influence of content-specific expectations on low-level orientation judgements

“…In both samples, there was no evidence for a relationship between the strength of content predictions and hallucination-proneness. This is in line with previous studies where content cues were not associated with hallucination-like perception (Haarsma et al, 2023). In contrast, in the discovery sample we did find that presence predictions have an enhanced influence on confidence in stimulus presence on absent trials, suggesting that presence priors are indeed overly strong.…”

“…The lack of an overly strong prediction effect in people prone to hallucinations in the replication sample seemingly contradicts the past literature. Although some past studies have found this type of effect in similarly sized or smaller, normative samples (Haarsma et al, 2023; Stuke et al, 2021), larger sample sizes may be required to observe these effects consistently. Furthermore, it is important to note that the average CAPS scores in this study were considerably lower (averaging around 45, in line with the original study by Bell et al, 2005) compared to some previous studies, which reported average CAPS scores of 107 (Stuke et al, 2021), 72 (Haarsma et al, 2023) and 58 (Schmack et al, 2021).…”

“…Instead, the content of false percepts arose from activity in the feedforward layers of the early visual cortex, independently of the cues signalling the most likely stimulus. However, implicit beliefs about stimulus presence rather than content could still have contributed to these false percepts (Haarsma et al, 2023). This raises the possibility that predictions of stimulus presence and content could have distinct effects on perceptual inference.…”

“…The predictive processing (PP) framework postulates that this process is inherent to all of perception such that what we perceive represents a combination of predictions, elicited by cues for example, and incoming sensory information (Bastos et al, 2012; Friston, 2005; Hohwy & Seth, 2020). Indeed, empirical research in recent decades suggests that predictions can bias perception towards the expected outcome through top-down connections that descend along the neural processing hierarchy (Aitken, Turner & Kok, 2020; Brandman & Peelen, 2017; Chalk et al, 2010; de Lange, Heilbron & Kok, 2018; Haarsma et al, 2022; Haarsma et al, 2023; Kok et al, 2013; Summerfield & de Lange, 2014; Thomas et al, 2022; Wyart, Nobre & Summerfield, 2012; Yon et al, 2018).…”

Bayesian Accounts of Perceptual Decisions in the Nonclinical Continuum of Psychosis: Greater Imprecision in Both Top-down and Bottom-up Processes

“…Layer specific neuroimaging techniques would be required for empirical verification, providing a promising avenue for future research (Haarsma, Kok, et al, 2022). Interestingly, recent research found that high confidence false percepts (a measure of hallucinatory propensity) were related to stimulus-like activity in middle input layers of the visual cortex in healthy participants (Haarsma, Deveci, et al, 2022). Task measures of hallucination propensity were also found to be associated with everyday hallucination severity, suggesting that hallucinatory-like perception may arise from spontaneous bottom-up activity in input layers of the visual cortex.…”