Theories of cognition seek to present models of the human mind that enlighten our understanding of a wide spectrum of cognitive processes including integration of sensory information, generation of behavior, and processes of high-level thought. Cognitive architectures in particular aim at providing computational instantiations of such models that facilitate the exploration of dynamic properties and behaviors. Prominent examples of such architectures are ACT-R [2], SOAR [3], CLARION [4], LIDA [5], or the highly influential Global Workspace Theory (GWT) by Bernard Baars [6].
In [1] we discuss two basic assumptions that commonly underlie theories of cognition that seem to be true almost trivially but may impact the neurobiological plausibility of these theories severely. The first assumption concerns the concept of memory, the second assumption concerns the related concept of computation. We argue that the concept of memory as it is commonly used in theories of cognition is strongly influenced by our understanding of memory as it is used in typical computer systems with von Neumann architecture. It is assumed that memory is a kind of container that allows to store pieces of information and to retrieve these pieces again later on. Similarly, the second assumption that we would like to challenge concerns the concept of computation. It is the idea that pieces of information present in such a memory can be used by different parts of a cognitive system, which implies an encoding of information whose meaning is shared by these different parts. We hope to show that both assumptions, when scrutinized, are difficult to reconcile with the properties and constraints inherent to a plausible neurobiological substrate. As an alternative, we present a different perspective that views memory as a process that is an intrinsic, distributed part of a neurobiological system.
References
1
,
Challenging the Intuition About Memory and Computation in Theories of Cognition,
In: Proceedings of the 11th International Joint Conference on Computational Intelligence - Volume 1: NCTA, (IJCCI 2019). INSTICC. SciTePress, pp. 522–527, 2019,
[pdf|doi|bibtex]
2
,
Act: A simple theory of complex cognition,
In: American Psychologist, (51):355–365, 1996,
[doi]
3
,
Soar: An architecture for general intelligence,
In: Artificial Intelligence, 33(1):1 – 64, 1987,
[doi]
4
,
The importance of cognitive architectures: an analysis based on clarion,
In: Journal of Experimental & Theoretical Artificial Intelligence, 19(2):159–193, 2007,
[doi]
5
,
Lida: A systems-level architecture for cognition, emotion, and learning,
In: IEEE Transactions on Autonomous Mental Development, 6, 2013,
[doi]
6
,
The Global Workspace Theory of Consciousness: Predictions and Results,
In: The Blackwell Companion to Consciousness, 2017,
[doi]