Predictive processing and hypnosis

Graham Jamieson
University of New England

Keywords: control, expectation, perception, prediction, predictive coding

Predictive Processing (PP) models are grounded in the insight of Boltzmann that the operating principle of the nervous system is to regulate its internal states to minimise surprise/ uncertainty. It does so though internal and external actions (changes) which ceaselessly fit and refit the expectations (predictions) generated within the system to match its own dynamically changing states. For example, the ambiguity in a pattern of light appearing suddenly on the periphery of the retina may be decreased through perceiving a twisted fallen branch or by leaping away and a surge of adrenalin.

In the case of perception, a ladder of exchanges between bottom-up signalling, driven by sensory receptors and top-down anticipation of lower-level inputs based on past learning interact to revise higher level anticipations to match (more or less) the latest change of inputs. The system transitions from one temporary stable coalition of predictions (generative model) to another driven by the reduction of uncertainty/entropy. The highest-level coherent set of predictions corresponds to the moment-by-moment perception-action-motivation states of the organism (self) (Pezzulo et al., 2024).

Within PP architectures, the mismatch between predictions and inputs to lower levels may drive the revision of predictions to reduce mismatch (predictive coding; PC), a process which dominates in perception, or predictions/expectations may drive effector systems to alter states of the internal (homeostasis) or external (action) environment to obtain a closer fit to anticipated goal states (active inference; AI). Whether PC or AI dominate at any moment depends upon the relative gain of predictions versus prediction errors, itself influenced by many factors (Yon & Frith, 2021).

Hypnotic suggestions pose counterfactual states of affairs which require responses at multiple levels of the self as if those events were real. Hypnotic responses necessarily juxtapose responses to consensus (objective) and imaginal (subjective) senses of inner and outer reality that are in shifting degrees aligned or counterpoised. PP provides a single unified framework in which the relative dominance of objective and subjective reality may be understood as the interplay between the gain of predictions and prediction errors, AI and PC in the dynamics of the self-system. The combined factors that influence these gains and the shifting balance of these factors constitute the major determinants of differences in response to hypnotic suggestion. The identification and utilisation of those factors then becomes a major focus for hypnosis research and application (Jamieson, 2016, 2018).

A version of PC, the comparator model of motor control, was successfully applied to predict and observe changes in neural correlates of the experience of volitional control (and/or bodily self-perception) in ideomotor and paralysis suggestions. In this model, a growing mismatch between predicted states of the body and somatosensory feedback linked to these suggestions underpinned the experience of action (or paralysis) as alien or not-self (Blakemore et al, 2003). Recently, pure AI models of action have proposed an alternative PP account leading to divergent theoretical predictions which remain to be tested (Jamieson, 2021; Martin & Pacherie, 2019).

For those theories of hypnosis which emphasise the role of (suggestions for) response expectancies and of top-down control processes in the implementation of responses to suggestion, generative models in conjunction with AI provide a unifying framework within which to model these processes (Lynn et al., 2023). Within the same framework, PP is also able to model dissociative phenomena in hypnosis (Zahedi et al., 2024) as dynamic shifts in the interplay between bottom-up and top-down control.

References
Blakemore, S. J., Oakley, D. A., & Frith, C. D. (2003). Delusions of alien control in the normal brain. Neuropsychologia, 41(8), 1058-1067. https://doi.org/10.1016/s0028-3932(02)00313-5

Jamieson, G. A. (2016). A unified theory of hypnosis and meditation states: The interoceptive predictive coding approach. In A. Raz, & M. Lifshitz (Eds.), Hypnosis and meditation: Towards an integrative science of conscious planes (pp. 313–342). Oxford University Press.

Jamieson, G. A. (2018). Expectancies of the future in hypnotic suggestions. Psychology of Consciousness: Theory, Research, and Practice, 5(3), 258–277. https://doi.org/10.1037/cns0000170

Jamieson, G. A. (2021). An insula view of predictive processing in hypnotic responses. Psychology of Consciousness: Theory, Research, and Practice, 9(2), 117-129. 10.1037/ cns0000266

Lynn, S. J., Green, J. P., Zahedi, A., & Apelian, C. (2023). The response set theory of hypnosis reconsidered: toward an integrative model. American Journal of Clinical Hypnosis, 65(3), 186-210. https://doi.org/10.1080/00029157.2022.2117680

Martin, J. R., & Pacherie, E. (2019). Alterations of agency in hypnosis: A new predictive coding model. Psychological review, 126(1), 133-152. DOI:10.1037/rev0000134

Pezzulo, G., Parr, T., & Friston, K. (2024). Active inference as a theory of sentient behavior. Biological Psychology, 108741. https://doi.org/10.1016/j.biopsycho.2023.108741

Yon, D., & Frith, C. D. (2021). Precision and the Bayesian brain. Current Biology, 31(17), R1026-R1032. https://doi.org/10.1016/j.cub.2021.07.044

Zahedi, A., Lynn, S. J., & Sommer, W. (2024). Cognitive simulation along with neural adaptation explain effects of suggestions: a novel theoretical framework. Frontiers in Psychology, 15, 1388347. doi.org/10.3389/fpsyg.2024.1388347