A Hybrid Account of Concepts Within the Predictive Processing Paradigm

Abstract

We seem to learn and use concepts in a variety of heterogenous “formats”, including exemplars, prototypes, and theories. Different strategies have been proposed to account for this diversity. Hybridists consider instances in different formats to be instances of a single concept. Pluralists think that each instance in a different format is a different concept. Eliminativists deny that the different instances in different formats pertain to a scientifically fruitful kind and recommend eliminating the notion of a “concept” entirely. In recent years, hybridism has received the most attention and support. However, we are still lacking a cognitive-computational model for concept representation and processing that would underpin hybridism. The aim of this paper is to advance the understanding of concepts by grounding hybridism in a neuroscientific model within the Predictive Processing framework. In the suggested view, the different formats are not distinct parts of a concept but arise from different ways of processing a functionally unified representational structure.

Grounded Cognition Entails Linguistic Relativity: Response to Commentators

“…This theory has been gaining prominence across the cognitive sciences, and it maintains that perception is basically the brain's “best guess” regarding the causes of its inputs, reflecting top‐down knowledge‐driven inference more than bottom‐up sensory‐driven feature processing (Clark, 2023). According to Slivac and Flecken (2023, p. 659), the concepts encoded by words fit into this framework because they “function as long‐term priors similar to other types of priors that shape perception, and that develop throughout infancy and childhood as our experiences with the world grow.” Other researchers have made the same point (Barrett, 2017; Barsalou, 2009; Michel, 2022), but Slivac and Flecken's (2023, p. 659) unique contribution is to emphasize that “cross‐linguistic differences in semantic … categories lead to differences in the language‐induced priors that people rely on.”…”

“…According to Slivac and Flecken (2023, p. 659), the concepts encoded by words fit into this framework because they "function as long-term priors similar to other types of priors that shape perception, and that develop throughout infancy and childhood as our experiences with the world grow." Other researchers have made the same point (Barrett, 2017;Barsalou, 2009;Michel, 2022), but Slivac and Flecken's (2023, p. 659) unique contribution is to emphasize that "cross-linguistic differences in semantic … categories lead to differences in the language-induced priors that people rely on." Slivac and Flecken (2023) discuss some ways in which this connection between semantic typology and predictive coding theory has implications for linguistic relativity.…”

Conceptual engineering, predictive processing, and a new implementation problem

“… a We consider concepts to be individuated by a root‐node, from which other nodes, for example, feature nodes emanate (e.g., Michel, 2020, 2022). This allows us to distinguish between the change to an existing concept and the replacement of a concept. …”

Copredication and Complexity Revisited: Reply to Murphy's Reply

“…However, one could in principle present a way of spelling out an “underspecification” account that goes beyond the high level of description we find in the literature by offering lower levels of description (e.g., including a neuromechanistic one), or a new variant that might deal better with some problematic or unaddressed phenomena. In fact, our aim was to transfer an independently plausible model of concepts spelled out within the PP paradigm (see also Michel, 2022) to the phenomenon of copredication. In our model, word senses vary in abstraction but not in “specification.” So, one could argue that there is never any privileged “fully specified” sense at all and the notion of “specification” should be given up (which is—maybe—a novel theoretical position).…”