Resting-state fMRI Signals of Intelligent People Wander in a Larger Space

Abstract

Natural intelligence is one of the vastly explored research areas in cognitive science. Its evolution and manifestation through behavioral patterns in animal kingdom have been extensively investigated. Since early days of cognitive sciences, there have been considerable efforts to simulate intelligent behaviors through high-level cognitive models. In the framework of the computational theory of mind, production systems are top-down models which simulate intelligent behaviors by invoking their behavioral manifestations. These models describe an intelligent behavior as structured mental programming which decomposes a complex task into simpler independent parts, each one represented by a cognitive enclosure where attention is sequentially devoted, and finally the information obtained from all cognitive enclosures is integrated to accomplish the task. In this article, we investigate the relations between these models of intelligence and resting-state fMRI signals. Based on these models, we hypothesize that the capacity of distinct mental representations is the core feature of intelligent behaviors. Therefore, we reason that resting-state fMRI signals of intelligent individuals wander in a larger space and can be divided to more well-separated independent components. This may be interpreted as the functional equivalence of one of the most celebrated structural correlates of intelligence, its positive association with the total brain volume. In the general framework of topological data analysis, using a well-established non-linear dimensionality reduction method, we show that indeed resting-state fMRI signals of intelligent individuals occupy a larger space and can be divided to more well-separated components with less connections in the reduced two-dimensional space. To our knowledge, this is the first attempt to relate the functional space of resting-state fMRI signals with the behavioral signatures of the human intelligence.