Active Inference and Cooperative Communication: An Ecological Alternative to the Alignment View

We present and contrast two accounts of cooperative communication, both based on Active Inference, a framework that unifies biological and cognitive processes. The mental alignment account, defended in Vasil et al., takes the function of cooperative communication to be the alignment of the interlocutor's mental states, and cooperative communicative behavior to be driven by an evolutionarily selected adaptive prior belief favoring the selection of action policies that promote such an alignment. We argue that the mental alignment account should be rejected because it neglects the action-oriented nature of cooperative communication, which skews its view of the dynamics of communicative interaction. We introduce our own conception of cooperative communication, inspired by a more radical ecological interpretation of the active inference framework. Cooperative communication, on our ecological conception, serves to guide and constrain the dynamics of the cooperative interaction via the construction and restructuring of shared fields of affordances, in order to reach the local goals of the joint actions in which episodes of cooperative communication are embedded. We argue that our ecological conception provides a better theoretical standpoint to account for the action-oriented nature of cooperative communication in the active inference framework.

Prior Beliefs Modulate Projection

Beliefs about the world affect language processing and interpretation in several empirical domains. In two experiments, we tested whether subjective prior beliefs about the probability of utterance content modulate projection, that is, listeners’ inferences about speaker commitment to that content. We find that prior beliefs predict projection at both the group and the by-participant level: the higher the prior belief in a content, the more speakers are taken to be committed to it. This result motivates the integration of formal analyses of projection with cognitive theories of language understanding.

Active inference through whiskers

Rodents use whisking to probe actively their environment and to locate objects in space, hence providing a paradigmatic biological example of active sensing. Numerous studies show that the control of whisking has anticipatory aspects. For example, rodents target their whisker protraction to the distance at which they expect objects, rather than just reacting fast to contacts with unexpected objects. Here we characterize the anticipatory control of whisking in rodents as an active inference process. In this perspective, the rodent is endowed with a prior belief that it will touch something at the end of the whisker protraction, and it continuously modulates its whisking amplitude to minimize (proprioceptive and somatosensory) prediction errors arising from an unexpected whisker-object contact, or from a lack of an expected contact. We will use the model to qualitatively reproduce key empirical findings about the ways rodents modulate their whisker amplitude during exploration and the scanning of (expected or unexpected) objects. Furthermore, we will discuss how the components of active inference model can in principle map to the neurobiological circuits of rodent whisking.

Optimal Interval Division

This paper provides a simple unified analysis of optimal interval division problems. My primitive is a cell function that assigns a value to each subinterval (cell). Submodular cell functions conveniently imply the property of decreasing marginal returns. Also, for coarse decision problems, optimal cutoffs commonly increase as prior belief shifts upward. Its implications on language and efficient menus are discussed.

Mitigating belief projection in explainable artificial intelligence via Bayesian teaching

State-of-the-art deep-learning systems use decision rules that are challenging for humans to model. Explainable AI (XAI) attempts to improve human understanding but rarely accounts for how people typically reason about unfamiliar agents. We propose explicitly modelling the human explainee via Bayesian teaching, which evaluates explanations by how much they shift explainees’ inferences toward a desired goal. We assess Bayesian teaching in a binary image classification task across a variety of contexts. Absent intervention, participants predict that the AI’s classifications will match their own, but explanations generated by Bayesian teaching improve their ability to predict the AI’s judgements by moving them away from this prior belief. Bayesian teaching further allows each case to be broken down into sub-examples (here saliency maps). These sub-examples complement whole examples by improving error detection for familiar categories, whereas whole examples help predict correct AI judgements of unfamiliar cases.

A Bayesian-Rational Framework for Conspiracy Theories

In our present era of fractured politics, social media, and fake news, conspiracy theories are as prominent as ever. While conspiracy theories are often dismissed as pathological or irrational reasoning, belief in at least some conspiracy theories could arise from a Bayesian rational system that is merely wrong, rather than truly irrational. This paper lays out a framework for understanding how conspiracy theories could arise from rational Bayesian cognition, identifying four potential sources for conspiracy theory belief in a primarily rational framework: elevated prior belief in CTs, different likelihoods, missing non-conspiratorial explanations, and non-epistemic utilities.

The cultural evolution of epistemic practices: the case of divination

While a substantial literature in anthropology and comparative religion explores divination across diverse societies and back into history, little research has integrated the older ethnographic and historical work with recent insights on human learning, cultural transmission and cognitive science. Here we present evidence showing that divination practices are often best viewed as an epistemic technology, and formally model the scenarios under which individuals may over-estimate the efficacy of divination that contribute to its cultural omnipresence and historical persistence. We found that strong prior belief, under-reporting of negative evidence, and mis-inferring belief from behavior can all contribute to biased and inaccurate beliefs about the effectiveness of epistemic technologies. We finally suggest how scientific epistemology, as it emerged in the Western societies over the last few centuries, has influenced the importance and cultural centrality of divination practices.

Prior beliefs modulate projection

Beliefs about the world affect language processing and interpretation in several empirical domains. In two experiments, we tested whether subjective prior beliefs about the probability of utterance content modulate projection, that is, listeners’ inferences about speaker commitment to that content. We find that prior beliefs predict projection at both the group and the by-participant level: the higher the prior belief in a content, the more speakers are taken to be committed to it. This result motivates the integration of formal analyses of projection with cognitive theories of language understanding.

Constrained versus Unconstrained Rational Inattention

The rational inattention literature is split between two versions of the model: in one, mutual information of states and signals are bounded by a hard constraint, while, in the other, it appears as an additive term in the decision maker’s utility function. The resulting constrained and unconstrained maximization problems are closely related, but, nevertheless, their solutions differ in certain aspects. In particular, movements in the decision maker’s prior belief and utility function lead to opposite comparative statics conclusions.

Limit Points of Endogenous Misspecified Learning

We study how an agent learns from endogenous data when their prior belief is misspecified. We show that only uniform Berk–Nash equilibria can be long‐run outcomes, and that all uniformly strict Berk–Nash equilibria have an arbitrarily high probability of being the long‐run outcome for some initial beliefs. When the agent believes the outcome distribution is exogenous, every uniformly strict Berk–Nash equilibrium has positive probability of being the long‐run outcome for any initial belief. We generalize these results to settings where the agent observes a signal before acting.