This dissertation proposes that the frontotemporal network involved in the neural substrate of language processing is as a subset of a domain-general system mediated by predictive processing mechanisms. Instead of attributing independent functions to each of its patterns of activity, it redescribes them as deriving from differences in the performance of this system when attempting integrations. In particular, it proposes that dorsal-ventral interactions are behind all linguistic and cognitive functions, representing as a single stream the exchange of feedforward (prediction error) and feedback (prediction) information under differing degrees of estimated certainty. This exchange is in turn manifested across a rostrocaudal hierarchy ending in pars triangularis, which connects the current processing context with retrieved task sets for its more efficient, proactive processing, and is generally involved in cognitive control.
The proposed domain-general system produces different cross-network interactivity patterns based on the ongoing levels of expected and unexpected uncertainty, sustaining language and cognition. Evidence for these is gathered from event-related potentials, oscillatory activity, neurochemistry, network neuroscience, and a broad range of cognitive neuroscience studies.
The last part of the dissertation studies the related problem of how language and speech could evolve simultaneously as a result of continuous, quantitative changes taking place in the aforementioned domain-general system. This section proposes a new framework to study the evolution of vocal learning abilities in connection with domain-general cognition, based on primate neural scaling rules and connectivity enhancements, and argues against comparative perspectives that only focus on animal vocalizations in order to explain human specificity.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados