Cognitive function synthesis: preliminary results

Abstract

Purpose This paper aims to introduce the cognitive function synthesis (CFS) conceptual framework to artificial general intelligence. CFS posits that at the “core” of intelligence in hybrid architectures, “interdependent” cognitive functions are synthesised through the interaction of various associative memory (AM)-based systems. This synthesis could form an interface layer between deliberative/symbolic and reactive/sub-symbolic layers in hybrid cognitive architectures. Design/methodology/approach A CFS conceptual framework, specifying an arrangement of AMs, was presented. The framework was executed using sparse distributed memory. Experiments were performed to investigate CFS autonomous extraction, consciousness and imagination. Findings Autonomous extraction was achieved using data from a Wi-Fi camera with the CFS auto-associative AM handling “Sensor Data”. However, noise reduction degraded the extracted image. An environment, simulated in V-REP 3.3.1, was used to investigate consciousness and imagination. CFS displayed consciousness by successfully tracking/anticipating the object position with over 90 per cent congruence. CFS imagination was seen by its predicting two time steps into the future. Originality/value Preliminary results demonstrate the plausibility of CFS claims for autonomous extraction, consciousness and imagination.

Validation of a Simulated Environment Developed for Validating CFS Autonom

“…Pounder, Ellis, Fernandez-Lopez [14] introduced the Cognitive Function Synthesis (CFS) conceptual framework, c.f. Fig.…”

“…CFS performance when governing the simulated robot platform would need to be compared with robot performance when operating under: a) 'Contact/Collision' reflex/ behaviour only, in the absence of any Obstacle Avoidance algorithm; and b) the influence of the Braitenberg Obstacle Avoidance algorithm. Further research would be required to treat with the demands for more sophisticated behaviours and behaviour coordination, as previously highlighted [14].…”

Description of a Simulated Environment Developed for Validating CFS Autonomy

“…1a). However, Pounder, Ellis, Fernandez-Lopez [1] note that, as a consequence of the Traditional AI Paradigm, integration challenges arise at all levels of AI, in general and in Artificial General Intelligence, or AGI, Hybrids in particular. In the latter architectures, cognitive functionality is introduced as independently subsisting cognitive modules at the symbolic and sub-symbolic levels, cf.…”

“…2 was developed from which the cognitive functions of: i) Perceptual Learning or Autonomous Extraction; ii) Consciousness; iii) Imagination; iv) Emotions; and v) Autonomous Action emerge. Pounder, Ellis, Fernandez-Lopez [1] performed experiments to investigate CFS display of functions i) to iii). Preliminary results suggest that the emergent properties, i.e.…”

“…Pounder, Ellis, Fernandez-Lopez [1] introduced the Cognitive Function Synthesis (CFS) conceptual framework to artificial general intelligence. CFS postulates that at the "core" of intelligence in hybrid architectures, "interdependent" cognitive functions are synthesised through the interaction of various associative memory (AM)-based systems.…”