Generative models, linguistic communication and active inference

Friston, Karl J.; Parr, Thomas; Yufik, Yan M.; Sajid, Noor; Price, Cathy J.; Holmes, Emma

doi:10.1016/j.neubiorev.2020.07.005

Cited by 75 publications

(66 citation statements)

References 133 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…linking different patterns of INS between individuals with different cognitive processes of interpersonal verbal communication, rather than from the computational, pure cognitive or linguistic perspectives. The latter two perspectives have also been discussed elsewhere ( Pickering and Garrod, 2004 2013 ; Hasson et al , 2012 ; Fusaroli et al , 2014 ; Friston and Frith, 2015a 2015b ; Pickering and Gambi, 2018 ; Friston et al , 2020 ). Finally, we conclude this article by highlighting some inspiring future directions in the social and cognitive neuroscience fields.…”

Section: Introductionmentioning

confidence: 97%

A hierarchical model for interpersonal verbal communication

Jiang

Zheng

2020

Social Cognitive and Affective Neuroscience

View full text Add to dashboard Cite

The ability to use language makes us human. For decades, researchers have been racking their minds to understand the relation between language and the human brain. Nevertheless, most previous neuroscientific research has investigated this issue from a “single-brain” perspective, thus neglecting the nature of interpersonal communication through language. With the development of modern hyperscanning techniques, researchers have begun probing the neurocognitive processes underlying interpersonal verbal communication and have examined the involvement of interpersonal neural synchronization (INS) in communication. However, in most cases, the neurocognitive processes underlying INS are obscure. To tentatively address this issue, we propose herein a hierarchical model based on findings from a growing amount of hyperscanning research. We suggest that three levels of neurocognitive processes are primarily involved in interpersonal verbal communication and are closely associated with distinctive patterns of INS. Different levels of these processes modulate each other bidirectionally. Furthermore, we argued that two processes (shared representation and interpersonal predictive coding) might coexist and work together at each level to facilitate successful interpersonal verbal communication. We hope this model will inspire further innovative research in several directions within the fields of social and cognitive neuroscience.

show abstract

Section: Introductionmentioning

confidence: 97%

A hierarchical model for interpersonal verbal communication

Jiang

Zheng

2020

Social Cognitive and Affective Neuroscience

View full text Add to dashboard Cite

show abstract

“…The discovery of mirror neurons inspired hopes that understanding of the origins of language can be "within our grasp" [161] Mirror neurons discharge during active movements of the hand or mouth (or both) performed by the subject or observed being performed by others (hence, the mirror neurons). It was hypothesized that the latter feature establishes a bridge from 'doing' to 'communicating' , or from acting to message sending [161,162]. Other hypothesis concerning language origins attribute its emergence to internal, as opposed to communicative, functions [35][36][37] and conceptualize language mechanisms as the manipulation of neuronal assemblies [163,164].…”

Section: Understanding and Languagementioning

confidence: 99%

Brain Functional Architecture and Human Understanding

Yufik¹

2021

Connectivity and Functional Specialization in the Brain

View full text Add to dashboard Cite

The opening line in Aristotle’s Metaphysics asserts that “humans desire to understand”, establishing understanding as the defining characteristic of the human mind and human species. What is understanding and what role does it play in cognition, what advantages does it confer, what brain mechanisms are involved? The Webster’s Dictionary defines understanding as “apprehending general relations in a multitude of particulars.” A proposal discussed in this chapter defines understanding as a form of active inference in self-adaptive systems seeking to expand their inference domains while minimizing metabolic costs incurred in the expansions. Under the same proposal, understanding is viewed as an advanced adaptive mechanism involving self-directed construction of mental models establishing relations between domain entities. Understanding complements learning and serves to overcome the inertia of learned behavior when conditions are unfamiliar or deviate from those experienced in the past. While learning is common across all animals, understanding is unique to the human species. This chapter will unpack these notions, focusing on different facets of understanding. The proposal formulates hypotheses regarding the underlying neuronal mechanisms, attempting to assess their plausibility and reconcile them with the recent ideas and findings concerning brain functional architecture.

show abstract

“…This is in line with the predictive processing view. Friston et al (2020) propose that the neuronal correlates of language processing and functional brain architectures should emerge naturally, given the right kind of generative model. The basic issue is that language processing is not in fact rule based, it is based in statistical correlations (see Section 5).…”

Section: The Poverty Of Stimulus Argumentmentioning

confidence: 99%

Neuroscience and Literacy: An Integrative View

Ellis

Bloch

2021

Preprint

View full text Add to dashboard Cite

Significant challenges exist globally regarding literacy teaching and learning, particularly in poor socio-economic settings in countries of the Global South. In this paper we argue that to address these challenges, major features of how the brain works that are currently ignored in the educational literature should be taken into account. First, perception is an active process based in detection of errors in hierarchical predictions of sensory data and action outcomes. Reading is a particular case. Second, emotions play a key role in underlying cognitive functioning. Innate affective systems underlie and shape all brain functioning, including oral and written forms of language and sign. Third, there is not the fundamental difference between listening/speaking and reading/writing often alleged on the basis of evolutionary arguments. Both are socio-cultural practices driven and learnt by the communication imperative of the social brain. Fourth, like listening, reading is not a linear, bottom-up process. Both are non-linear contextually shaped psycho-social processes of understanding, shaped by current knowledge and cultural contexts and practices. Reductionist neuroscience studies which focus on decontextualized parts of reading cannot access all the relevant processes. An integrated view of brain function reflecting this non-linear nature implies that an ongoing focus on personal meaning and understanding provides positive conditions for all aspects of literacy learning. Assessment of literacy teaching at all its stages should include indicators that take into account these foundational features relating reading and writing to neuroscience.

show abstract

Generative models, linguistic communication and active inference

Abstract: Highlights New (hierarchical) generative model for linguistic interactions. Builds on active inference formulations of dyadic interactions. We simulate agents that ask and answer questions together. Theta-gamma coupling emerges from belief updating under this framework.

Cited by 75 publications

References 133 publications

A hierarchical model for interpersonal verbal communication

A hierarchical model for interpersonal verbal communication

Brain Functional Architecture and Human Understanding

Neuroscience and Literacy: An Integrative View

Contact Info

Product

Resources

About