ReaCog, a Minimal Cognitive Controller Based on Recruitment of Reactive Systems

Abstract

It has often been stated that for a neuronal system to become a cognitive one, it has to be large enough. In contrast, we argue that a basic property of a cognitive system, namely the ability to plan ahead, can already be fulfilled by small neuronal systems. As a proof of concept, we propose an artificial neural network, termed reaCog, that, first, is able to deal with a specific domain of behavior (six-legged-walking). Second, we show how a minor expansion of this system enables the system to plan ahead and deploy existing behavioral elements in novel contexts in order to solve current problems. To this end, the system invents new solutions that are not possible for the reactive network. Rather these solutions result from new combinations of given memory elements. This faculty does not rely on a dedicated system being more or less independent of the reactive basis, but results from exploitation of the reactive basis by recruiting the lower-level control structures in a way that motor planning becomes possible as an internal simulation relying on internal representation being grounded in embodied experiences.

From Adaptive Locomotion to Predictive Action Selection – Cognitive Control for a Six-Legged Walker

“…Such a structure has already been developed and tested with a simulated agent as a proof of concept [39] and could show the feasibility of the general concept of internal simulation to find a behavior as a solution to a novel problem. In that example, the agent was placed in simulation into an unstable position that could not be resolved by the reactive controller.…”

“…It is motivated by broad behavioral and neuroscientific support for flexible recruitment of internal models in cognitive tasks in other (higher) animals including humans [50]- [52]. The cognitive expansion consists of different parts: First, an adaptation of the global control network to detect problems and to run internal simulations [39] (see Fig. 3, top left).…”

Integrative Biomimetics of Autonomous Hexapedal Locomotion

“…When decoupling the body model from the actual joint drives, the same dynamic internal body model may also be used for movement prediction and planning. Recently, we applied it as an internal simulator to forecast the consequences of different alternative behaviors as a form of planning ahead (Schilling and Cruse, 2017). In this series of simulations, the model served a dual purpose, exploiting its full flexibility in motor control and planning.…”

“…These estimates allowed to decide whether the chosen behavior would lead to instability or else might help to overcome the problematic situation. Only if the internal simulation proved successful, the internally simulated behavior was applied to the actuators of the system (Schilling and Cruse, 2017). This shows how an embodied internal model may be grounded in lower-level motor control and can be used flexibly for a cognitive task such as planning ahead (Cruse and Schilling, 2016).…”

Adaptive Control Strategies for Interlimb Coordination in Legged Robots: A Review

“…By using this control technique, versatile behaviors (e.g., gap crossing, obstacle crossing, and global planning to avoid or attack obstacles) can be generated to deal with complex environments (Figure 5C ). Furthermore, Schilling and Cruse ( 2017 ) expanded Walknet to invent new behaviors and test them by internal simulation before using them in reality. Arena et al ( 2017 ) proposed multilayered CPG-based locomotion control with insect inspired motor-skill learning.…”