How brains build higher order representations of uncertainty

Higher-order representations are neural or computational states that are “about” first-order representations, encoding information not about the external world per se but about the agent’s own representational processes – such as the reliability, source, or structure of a first-order representation. These higher-order representations appear critical to metacognition, learning, and even consciousness by some accounts, yet their dimensionality, construction, and neural substrates remain poorly understood. Here, we propose that metacognitive estimates of uncertainty or noise reflect a read-out of higher-order “posteriors” from a Bayesian perspective. We then discuss how these higher-order posteriors reflect a combination of higher-order “likelihoods” (current uncertainty evidence, conditioned on true first-order uncertainty) and “priors” – learned distributions over expected first-order uncertainty – and how various emerging engineering and theory-based analytical approaches may be employed to examine the estimation processes and neural correlates associated with these highly under-explored components of our experienced uncertainty.