Bridging Connectionism and Relational Cognition through Bi-directional Affective-Associative Processing

Abstract

Connectionist architectures constitute a popular method for modelling animal associative learning processes in order to glean insights into the formation of cognitive capacities. Such approaches (based on purely feedforward activity) are considered limited in their ability to capture relational cognitive capacities. Pavlovian learning value-based models, being not based purely on fully connected feedforward structure, have demonstrated learning capabilities that often mimic those of ‘higher’ relational cognition. Capturing data using such models often reveals how associative mechanisms can exploit structure in the experimental setting, so that ‘explicit’ relational cognitive capacities are not, in fact, required. On the other hand, models of relational cognition, implemented as neural networks, permit formation and retrieval of relational representations of varying levels of complexity. The flexible processing capacities of such models are, however, are subject to constraints as to how offline relational versus online (real-time, real-world) processing may be mediated. In the current article, we review the potential for building a connectionist-relational cognitive architecture with reference to the representational rank view of cognitive capacity put forward by Halford et al. Through interfacing system 1-like (connectionist/associative learning) and system 2-like (relational-cognition) computations through a bidirectional affective processing approach, continuity between Halford et al’s cognitive systems may be operationalized according to real world/online constraints. By addressing i) and ii) in this manner, this paper puts forward a testable unifying framework for system 1-like and system 2-like cognition.

Feeling and thought in collective action on social issues: Toward a systems perspective

“…We argue that many problems in the psychological study of action could be alleviated by increased engagement with a systems meta‐theoretical approach, which has long been advocated for in psychology and greatly benefitted other areas (e.g., Bou Zeineddine & Pratto, 2017; Leach & Bou Zeineddine, in press‐a; Lowe et al., 2019; Markus & Hamedani, 2007; Meagher, 2020; Nowak et al., 2017; Pessoa, 2019; Schill et al., 2019; Sosnowska et al., 2020; Witherington, 2007). This meta‐theory suggests that (human) agents and environments co‐evolve as a unitary system and that this co‐evolution produces a self‐organized emergent order, that is, unstable and nonlinear in its change (e.g., Eidelson, 1997).…”

Studying Consensus and Disagreement during Problem Solving in Teams through Learning and Response Generation Agents Model

“…New ideas are generated in the LRGA model using either a fast, top-down flow or a slower, bottom-up flow guided by parameters grounded in psychology, like affective valuation, to determine the nature of emotions based on the differences between expected outcomes and received inputs [16,24,49], responses to asymmetrical utilities [58], dominant concepts that capture the agent's attention [38], minimizing the required effort [58], or maximizing the accessed knowledge. Idea creation through the slower, bottom-up flow is similar to the five design strategies discussed in ref.…”

“…Common ground knowledge, shared visual information, and beliefs about the other team members influence postural and gaze coordination [19]. Affective valuation is also important during interaction [16,24], such as different interpretations of the external stimuli and distinct expectations for the outputs of problem solving. Members must be committed to participating in the team effort.…”

“…The above cases refer to participants that interact by stating their ideas generated in response to others' inputs and their own beliefs and experiences. Emotional and social cues are important in the process too [15,16]. The exchanged ideas are syllogisms [17].…”