Spatiotemporal Abstraction Theory: Re-Interpretation of Localized Cortical Networks
The brain excels at extracting meaning from noisy and degraded input, yet the computational principles that underlie this robustness remain unclear. We propose a theory of spatiotemporal abstraction (STA), in which localized cortical networks integrate inputs across space and time to produce multi-scale, concept-level representations that remain stable despite loss of detail. We demonstrate how this principle explains a long-standing paradox of how cochlear implant patients can understand speech despite severely scrambled neural patterns. STA provides a unified framework that explains fundamental questions: Why do we have so many neurons that respond very similarly in one cortical location? Why do we have different inhibitory neurons? It also forces us to re-examine long-standing explanations of memory, creativity, illusions, attractor dynamics, excitatory-to-inhibitory balance, and the structure and purpose of the ubiquitous canonical circuits seen throughout the brain. We conclude with STA implications for improving neural implants and artificial neural networks.