Abstract
According to the ‘Bayesian brain’ or ‘predictive coding’ accounts, the brain makes sense of incoming sensory information based on three processes: 1) sampling of sensory evidence, 2) extraction of statistical regularities to create ‘predictive models’ about the environment, and 3) exploitation of such predictive models. These components interact dynamically to optimize the perceptual decision-making process and to guarantee flexibility to changes. The available evidence about the role of predictive coding in perception is mostly behavioral and only few studies investigated its neurophysiological basis. Importantly, there have been virtually no attempts to probe whether the key components of predictive coding could be modulated non-invasively in the human brain to promote faster learning and neural plasticity. The present project is the first systematic study of the plasticity of neural networks underlying predictive processes in the context of human perception. The project used techniques that aimed to: i) modulate the precision of sensory sampling; ii) modulate the “learning” and iii) “exploitation” of statistical regularities, and thus the ability to generate predictive models. We employed different and complementary non-invasive brain stimulation (NIBS) approaches across the involved units (UNIBO, UNIPD, UNITN). First, transcranial alternating current stimulation (tACS) has been used to modulate specific brain frequencies. Occipital tACS in the alpha range (8-14 Hz) has been employed to test whether the rate of temporal sampling of visual information can be modulated. Frontal tACS within the theta band (4-7 Hz) has been used to test whether the learning of statistical regularities can be improved and whether this affects the exploitation of predictive models in new settings. Second, the project used transcranial random noise stimulation (tRNS), given its potential to boost plasticity in sensory and associative areas. tRNS in occipital areas was used to modulate the signal-to-noise ratio of sensory sampling. tRNS in frontal areas has been used to modulate learning of priors from statistical regularities. Finally, we used a neurostimulation paradigm named ccPAS (cortico-cortical Paired Associative Stimulation), which can modify the connectivity of brain regions by inducing Hebbian-like plastic changes. Occipito-occipital ccPAS has been employed to strengthen re-entrant (feedback) processing and test its impact on visual sensory sampling. Parieto-occipital ccPAS has been used to strengthen the connectivity between cortical areas important for prior model exploitations. The project tested healthy human volunteers to provide important basic knowledge on the plasticity of the predictive brain. Since predictive processes have been hypothesized to be altered in different disorders, potential extensions could include the investigation of predictive processes dynamics in relation to individual differences in sensory sampling, cognitive styles and personality traits.
Results achieved
The project carried out the first systematic study of the plasticity of the neural networks underlying predictive processes in human perception, using non-invasive brain stimulation (NIBS) to modulate the three components of predictive coding: the sampling of sensory evidence, the learning of statistical regularities, and the exploitation of the predictive models thus generated. Within the UNIBO, work focused on Cortico-Cortical Paired Associative Stimulation (ccPAS), a paradigm able to reshape connectivity between cortical areas through Hebbian-like plasticity. Protocols were used to strengthen connectivity along functionally distinct directions of the visual pathway: V5→V1 ccPAS, aimed at boosting feedback (re-entrant) processing, and IPS→V1 ccPAS, aimed at strengthening the parieto-occipital connectivity relevant for prior exploitation; the reverse directions (V1→V5 and V1→IPS) were also collected. Combining behavioural and EEG measures, pre-stimulus individual alpha peak frequency (IAF), extracted from left-occipital channels, was used as a neural marker of perceptual sampling. Mixed-model analyses showed, after stimulation, a reduction in accuracy in both groups, particularly evident for one of the two cues, and a reduction in response bias also present in both groups. The picture changed, however, once pre-stimulus IAF was taken into account: the bias reduction vanished in the IPS-V1 group, while it persisted in the V5-V1 group, where it was in fact enhanced by higher IAF. Consistently, IAF decreased between the learning and testing phases only in the IPS-V1 group, remaining stable in the V5-V1 group; moreover, only in the latter did correct responses tend to be associated with higher IAF. Together, these results suggest that V5→V1 ccPAS strengthens the role of IAF in perceptual sampling and may counteract the decline observed after IPS→V1 stimulation, providing causal evidence for the contribution of feedback connectivity and for IAF as a temporally resolved marker of perceptual inference. The results and the theoretical framework of the project have fed into several international publications: – Tarasi L, Turrini S, Sel A, Avenanti A, Romei V (2024). Cortico-Cortical Paired Associative Stimulation (ccPAS) to Investigate the Plasticity of Cortico-Cortical Visual Networks in Human. Current Opinion in Behavioral Sciences. – Tarasi L, Tabarelli de Fatis C, Covelli M, Ippolito G, Avenanti A, Romei V (2025). Preparing to act follows Bayesian inference rules. iScience, 28(6), 112645. – Tarasi L, Romanazzi D, Pasini A, Romei V (2025). Delusion-like thinking is associated with lower individual alpha peak frequency. Schizophrenia, 11(1), 76. – Turrini S, Fiori F, Arcara G, Romei V, di Pellegrino G, Avenanti A (2025). State-dependent associative plasticity highlights function-specific premotor–motor pathways crucial for arbitrary visuomotor mapping. Science Advances, 11(20), eadu4098. The project, coordinated by the University of Trento in collaboration with the Bologna (P.I. Vincenzo Romei) and Padua units, provided fundamental knowledge on the plasticity of the predictive brain, with potential extensions to individual differences in sensory sampling, cognitive styles and personality traits.Project details
Unibo Team Leader: Vincenzo Romei
Unibo involved Department/s:
Dipartimento di Psicologia "Renzo Canestrari"
Coordinator:
Libera Università Vita Salute S.Raffaele MILANO(Italy)
Total Unibo Contribution: Euro (EUR) 91.641,00
Project Duration in months: 24
Start Date:
30/11/2023
End Date:
28/02/2026
