Thinking through prior bodies: autonomic uncertainty and interoceptive self-inference

Abstract

The Bayesian brain hypothesis, as formalized by the free-energy principle, is ascendant in cognitive science. But, how does the Bayesian brain obtain prior beliefs? Veissière and colleagues argue that sociocultural interaction is one important source. We offer a complementary model in which “interoceptive self-inference” guides the estimation of expected uncertainty both in ourselves and in our social conspecifics.

Associative learning and extinction of conditioned threat predictors across sensory modalities

“…While our results are limited by the lack of complete SCR data as a biological marker of learning as well as by a more definite exclusion of pain-modality-specific vascular artefacts induced by gasping or other respiratory or movement-related effects that could be more closely inspected whether pulse oximetry or respiration had been measured, they do provide a more refined understanding of conditioned nocebo effects in the context of clinically-relevant interoceptive and exteroceptive threats. Merging our clinically-driven perspective with the rapidly expanding general literature on interoception and predictive processing provides opportunities for the development or refinement of computational models based on the precision of interoceptive versus exteroceptive signals, advancing not only the definition of salience itself but also clarification its mechanistic basis 88 – 93 . The translation of this knowledge may ultimately help understand and minimise negative expectancy effects in patients with chronic pain 94 , 95 , especially in disorders of gut–brain interactions, adds a brain perspective to the eloquent claim that the gut is ‘smart’ due to its capability to learn and remember 96 , and supports further efforts towards extinction-based treatment approaches for these highly prevalent conditions 10 , 97 .…”

The heart rate discrimination task: a psychophysical method to estimate the accuracy and precision of interoceptive beliefs

“…Numerous computational accounts emphasize that accurate metacognitive self-monitoring is likely to depend on a process by which the precision of the sensory signals underlying the type 1 decision is “read-out” by a higher-order metacognitive module, such that noisy, imprecise signals can be expected to degrade both perceptual performance and metacognitive sensitivity (Fleming et al, 2012; Maniscalco & Lau, 2016). However, other accounts emphasize that top-down “self-beliefs” may play a crucial role in shaping the interaction between low-level precision and higher-order metacognition (Allen et al, 2020; Fleming & Daw, 2017). Speculatively, our findings suggest that in the cardiac domain, metacognition may be largely dominated by top-down beliefs, rather than pure sensory read-out.…”

What might interoceptive inference reveal about consciousness?

“…Interoceptive prediction errors and precision thereof are here thought to play a unique role in determining what is salient for an agent in any given context, such that unexpected challenges to homeostasis or allostasis essentially govern the innate value of different outcomes. Computational and conceptual models have expanded on this view to describe a process of metacognitive and interoceptive self-inference, in which the a priori expected precision afforded the homeostatic and allostatic fluctuations is always higher than that of over sensory-motor channels [24,64,65]. As fluctuations in, for example, blood temperature or arterial pulsation, can directly modulate the noise (i.e., inverse precision) of neuronal circuits in a global fashion, then the representation of expected precision is argued to both sample directly from the precision of interoceptive prediction errors, and to utilize descending visceromotor control as a means of optimizing sensory precision.…”

“…Generally, in these accounts, what achieves ignition can be understood as the relationship between the expected (top-down) and sensory (bottom-up) precision, where when the self-model increases the precision of lower order modules, they become better fitting models of the world and are more likely to reach ignition. Further work has proposed that interoception may play a crucial role within the self-model, by either conditioning expected precision [64,65], or by modulating the degree to which lower-order representations are interpreted as related to the sense of self [25,61,67,[82][83][84].…”

“…This, of course, is only feasible (i.e., free-energy minimizing) if the body and the controlling system (for example the brain) are so tightly coupled that distinguishing between them has only irrelevant advantages to the agent's resulting behaviour. In addition, given that maintenance of the body is crucial for the controlling system's self-maintenance, the pragmatic value -that is, prior preferences over outcomes -for an active inference agent should largely be defined in terms of -or at least in relation to -bodily states, and therefore interoception, largely in accordance with the idea of homeostatic priors and inference as having a privileged position as a first prior [36,65]. This clearly posits interoception, self-maintenance, emotion, interoceptive inference, value and consciousness as tightly interlinked concepts.…”