Abstract
Biological and artificial intelligence (AI) are often defined by their capacity to achieve a hierarchy of short-term and long-term goals that require incorporating information over time and space at both local and global scales. More advanced forms of this capacity involve the adaptive modulation of integration across scales, which resolve computational inefficiency and explore-exploit dilemmas at the same time. Research in neuroscience and AI have both made progress towards understanding architectures that achieve this. Insight into biological computations come from phenomena such as decision inertia, habit formation, information search, risky choices and foraging. Across these domains, the brain is equipped with mechanisms (such as the dorsal anterior cingulate and dorsolateral prefrontal cortex) that can represent and modulate across scales, both with top-down control processes and by local to global consolidation as information progresses from sensory to prefrontal areas. Paralleling these biological architectures, progress in AI is marked by innovations in dynamic multiscale modulation, moving from recurrent and convolutional neural networks—with fixed scalings—to attention, transformers, dynamic convolutions, and consciousness priors—which modulate scale to input and increase scale breadth. The use and development of these multiscale innovations in robotic agents, game AI, and natural language processing (NLP) are pushing the boundaries of AI achievements. By juxtaposing biological and artificial intelligence, the present work underscores the critical importance of multiscale processing to general intelligence, as well as highlighting innovations and differences between the future of biological and artificial intelligence.
Highlights
Recent work suggests that the brain’s computational capacities act over multiple spatial and temporal scales and that understanding this multiscale ‘attention’ is critical for explaining human behavior in complex real world environments [1,2,3,4,5,6]
The brain is equipped with mechanisms that can represent and modulate across scales, both with top-down control processes and by local to global consolidation as information progresses from sensory to prefrontal areas
How general are problems in multiscale processing across tasks and domains? What evidence is there that biological and artificial intelligence benefit from multiscale processing? What architectures best achieve multiscale processing in biological and artificial intelligence? How can these architectures move from fixed to adaptive multiscale intelligence? What environmental conditions benefit most from multiscale processing? how can we develop predictive behavioral models using a more comprehensive and generalizable multiscale theoretical framework of biological cognition and AI [16,17]? The requirement to integrate information over spatial and temporal scales in a wide variety of environments would seem to be a common feature underlying intelligent systems, and one whose performance has a profound impact on behavior [16,17,18,19,20,21]
Summary
Recent work suggests that the brain’s computational capacities act over multiple spatial and temporal scales and that understanding this multiscale ‘attention’ (broadly defined) is critical for explaining human behavior in complex real world environments [1,2,3,4,5,6]. Other examples of the generality of multiscale processing include developing general problem solving and information search strategies [31,32], designing versatile AI systems [9], planning efficient business management strategies [33], and establishing a comprehensive understanding of mental illness [34] In these challenging real world environments, adaptive agents have to evaluate the long-term costs of current solutions, find and evaluate alternative solutions, and vary the scale over which these solutions are evaluated in light of the current context [31]. A multiscale architecture that can effectively modulate between local tasks while considering multiple global goals and contextual factors is necessary for agents to perform well in dynamic and uncertain environments approaching real world complexity We believe that both fields stand to gain from a common understanding of this problem. We will describe two formal demonstrations of this narrowing of the temporal scale: decision inertia and habit formation
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call