Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

From Manifestations to Cognitive Architectures: a Scalable Framework

Alfredo Ibias, Guillem Ramirez-Miranda, Enric Guinovart, Eduard Alarcon

TL;DR

The paper tackles the gap between optimisation-focused AI and brain-inspired cognition by proposing a scalable framework that treats the world as an information source and builds cognitive structures from primitive Spatial Distributed Representations. It introduces Footprinting and a hierarchical Cell–Cluster–Metacluster architecture to enable cross-modal integration and sensorimotor loops, anchored by the Self-Projecting Persistence Principle (SPPP) that ensures representations persist and project themselves over time. This approach yields Footprints and Projections as cognitive primitives and culminates in a Synthetic Cognition composed of Motoperceptive, Declarative, and Procedural Metaclusters, offering a bottom-up path toward general intelligence. The framework aims to address longstanding questions such as Systematicity by enabling structured, hierarchical abstractions and reasoning capabilities with potential practical impact on scalable, brain-inspired AI systems.

Abstract

The Artificial Intelligence field is flooded with optimisation methods. In this paper, we change the focus to developing modelling methods with the aim of getting us closer to Artificial General Intelligence. To do so, we propose a novel way to interpret reality as an information source, that is later translated into a computational framework able to capture and represent such information. This framework is able to build elements of classical cognitive architectures, like Long Term Memory and Working Memory, starting from a simple primitive that only processes Spatial Distributed Representations. Moreover, it achieves such level of verticality in a seamless scalable hierarchical way.

From Manifestations to Cognitive Architectures: a Scalable Framework

TL;DR

The paper tackles the gap between optimisation-focused AI and brain-inspired cognition by proposing a scalable framework that treats the world as an information source and builds cognitive structures from primitive Spatial Distributed Representations. It introduces Footprinting and a hierarchical Cell–Cluster–Metacluster architecture to enable cross-modal integration and sensorimotor loops, anchored by the Self-Projecting Persistence Principle (SPPP) that ensures representations persist and project themselves over time. This approach yields Footprints and Projections as cognitive primitives and culminates in a Synthetic Cognition composed of Motoperceptive, Declarative, and Procedural Metaclusters, offering a bottom-up path toward general intelligence. The framework aims to address longstanding questions such as Systematicity by enabling structured, hierarchical abstractions and reasoning capabilities with potential practical impact on scalable, brain-inspired AI systems.

Abstract

The Artificial Intelligence field is flooded with optimisation methods. In this paper, we change the focus to developing modelling methods with the aim of getting us closer to Artificial General Intelligence. To do so, we propose a novel way to interpret reality as an information source, that is later translated into a computational framework able to capture and represent such information. This framework is able to build elements of classical cognitive architectures, like Long Term Memory and Working Memory, starting from a simple primitive that only processes Spatial Distributed Representations. Moreover, it achieves such level of verticality in a seamless scalable hierarchical way.
Paper Structure (5 sections, 3 figures)

This paper contains 5 sections, 3 figures.

Table of Contents

  1. Introduction
  2. A Computational Framework of Cognition
  3. Conclusions
  4. Acknowledgments.
  5. Disclosure of Interests.

Figures (3)

  • Figure 1: (left) An example of Footprint: the combination of the 1's of the MNIST dataset. (right) An example of Cell: the Footprints of the 60,000 samples of the MNIST dataset.
  • Figure 2: (left) An example of Cluster: the Cells of the hierarchy present in the 60,000 samples of the MNIST dataset. The seed Cell is the central black node. (right) An example of Metacluster: a Metacluster for an agent that has two sensor inputs (image and sound) and a motor output.
  • Figure 3: A generic example of Synthetic Cognition.