Chapter 3. Categories at the interface of cognition and action

Abstract

Grounded cognition offers a natural framework for studying categories that lie at the interface of cognition and action. From this perspective, cognition emerges from the coupling of the brain, the modalities, the body, and the environment. Situated action, in particular, links these domains together, as perceived entities and events in the world (e.g., hedonic stimuli, social agents) trigger self-relevant responses (e.g., goals, values, identities, norms), which in turn produce bodily states (e.g., affect, motivation) that initiate actions (e.g., bodily actions, vocalizations) and ultimately produce outcomes, again in the world (e.g., reward, punishment). Ad hoc and goal-derived categories emerge between the internal cognitive states and external physical states that arise during the pursuit of situated action. Although these categories behave similarly in some ways to conventional taxonomic categories (e.g., possessing graded structure), they differ in others, especially in their dependence on context (typically contexts of situated action). On many occasions, these categories are constructed to support current situated action (ad hoc categories), but when they become relevant to situated action across many occasions, they become wellestablished in memory (goal-derived categories). Across these categories, ideals play central roles in determining graded structure, supporting the goals that drive situated action. Event frames offer a natural means of understanding how ad hoc and goal-derived categories become constructed compositionally, and how they offer an interface between cognition and the world, essentially providing coordinated patterns of values for instantiating frame variables. Interestingly, this frame-based account can be naturally implemented in the simulation mechanisms of perceptual symbol systems, further grounding ad hoc categories in relations between the brain, modalities, body, and world during situated action.

Issues in Grounded Cognition and how to solve them – The Minimalist Account

“…It emphasizes more than its predecessors the role of language as a shortcut or pointer to concepts. This connects to the larger situated action cycle which describes how such modal simulations connect to action (Barsalou, 2020(Barsalou, , 2021.…”

Thinking and Learning in Nested Systems: The Individual Level

“…As elements of a situation are perceived, they are mapped to cognitive structures in the mind (concepts, knowledge schemas), which support inferential reasoning. A concept can be viewed as a locally coherent dynamic network of multimodal elements that support sense-making, prediction, and action in different contexts. , For example, in perceiving and categorizing an entity as an “apple” one implicitly associates attributes from several categories such as object, solid, round, edible, and fruit but also activates representations of internal states like hunger or pleasure, perceptual responses that create the sensation of a sour taste, and mental simulations of actions for how to eat the apple. The constructs that are activated depend on the context.…”

Grounding Context in Embodied Cognitive Robotics

“…Every scene could have multiple levels of structure, from the level of feature representations to individual items to the level of ensembles of objects. Each scene representation allows simulations regarding the activated context-specific category in support of situated action (Barsalou, 2020a ).…”