Can you predict intelligence? Neuroscientists figure out how

Now, a new study provides one of the most in-depth views of how different brain regions and neural networks affect a person's general intelligence, essentially their ability to solve problems in various situations.

For the study, scientists looked to the biological foundations of cognitive abilities within the brain, employing "connectome-based predictive modeling" in order to compare five theories describing how the brain generates intelligence.

Their goal was to understand how our ability to solve problems is made possible by the information-processing architecture of the brain, as explained by the study's lead researcher Aron Keith Barbey, professor of psychology, bioengineering, and neuroscience at the University of Illinois Urbana-Champaign in a press release.

The paper's first author, Evan Anderson, now a researcher for Ball Aerospace and Technologies Corp., elaborated that getting a deeper insight into the biology of cognitive abilities calls for "characterizing how individual differences in intelligence and problem-solving ability relate to the underlying architecture and neural mechanisms of brain networks."

A new approach

Theories of intelligence tend to utilize several approaches. Some look at the prefrontal cortex, which is important to cognitive processes that involve making decisions and solving problems, as well as other localized regions of the brain.

According to Barbey, recent theories zero in on particular brain networks, focusing on how they interact with other.

In their study, the researchers tested a number of current theories against the "network neuroscience theory," which Barbey and Anderson developed. This theory proposes that intelligence is the byproduct of the brain's global architecture and the interplay of both strong and weak connections. 

"Strong connections involve highly connected hubs of information-processing that are established when we learn about the world and become adept at solving familiar problems," Anderson shared in the press release.

"Weak connections have fewer neural linkages but enable flexibility and adaptive problem-solving." Working together, strong and weak connections "provide the network architecture that is necessary for solving the diverse problems we encounter in life."

Their study consisted of a diverse group of 297 undergrad students who completed a series of general intelligence tests that gauged their problem-solving abilities and how well they could adapt to different situations. Each participant then underwent an fMRI resting-state scan of their brain. 

The scientists wanted to capture the great variety of neural networks that are active even when we are resting. As Professor Barbey explained, these networks include the frontoparietal network, responsible for making goal-oriented decisions, the dorsal attention network that helps with visual and spatial awareness, as well as the salience network, used in focusing our attention on the most important stimuli.

Which theory 'won'? 

The researchers used the responses and data gathered from the students to see which theory better predicted how well the participants would do on tests of intelligence.

They found that their theory, which took into account the whole brain, achieved more accuracy in predicting a participant's ability to solve problems and adapt.

Theories that looked only at specific brain regions or networks did not do as well in several areas. Barbey believes this means that "global information processing" is paramount to a person's ability to do well in cognitive challenges. 

"Rather than originate from a specific region or network, intelligence appears to emerge from the global architecture of the brain and to reflect the efficiency and flexibility of systemwide network function," Barbey stated in the press release.

Interesting Engineering (IE) asked Professor Barbey for more details on their work. We asked how "network neuroscience theory" takes into account the overall architecture of the brain to predict or describe intelligence.

To answer this question, Barbey pointed to his review of the network neuroscience theory presented in a 2017 article for Trends in Cognitive Sciences, titled "Network Neuroscience Theory of Human Intelligence."

As he writes in that paper, network neuroscience is an emerging interdisciplinary field, which incorporates methods from mathematics, physics, and computer science "to enable the formal measurement and modeling of the interactions among network elements, thereby providing a powerful new lens for examining the emergence of global network phenomena."

"The paper also lays out why Network Neuroscience Theory is a new point of view on theories of intelligence,  with Barbey explaining that general intelligence originates from individual differences in "the system-wide topology and dynamics of the human brain."

According to Network Neuroscience Theory, "the small-world topology of brain networks enables the rapid reconfiguration of their modular community structure, creating globally coordinated mental representations of a desired goal-state and the sequence of operations required to achieve it."

Under this approach, individual differences in intelligence arise from how quickly someone's brain can exchange information between different networks. As such, intelligence can be considered a product of a person's global information processing capacity in order to engage "easy-to-reach network states to construct mental representations for crystallized intelligence based on prior knowledge and experience, and accessing difficult-to-reach network states to construct mental representations for fluid intelligence based on cognitive control functions that guide adaptive reasoning and problem-solving," writes Barbey.

Professor Barbey also shared some key points from a summary of the theory that he presented previously to American Mensa on whether there is a particular map or signature of intelligence revealed by the fMRI that can be adopted to predict the general intelligence of a person.

Contemporary research conceives of the brain as a dynamic and active inference generator that anticipates incoming sensory inputs, forming hypotheses about that world that can be tested against sensory signals that arrive in the brain. Plasticity is therefore critical for the emergence of human intelligence — providing a powerful mechanism for updating prior beliefs, generating dynamic predictions about the world, and adapting in response to ongoing changes in the environment.

We have developed a network neuroscience theory of general intelligence that is based on these ideas. According to the theory, general intelligence reflects individual differences in the efficiency and flexibility of brain networks. The human brain is designed for efficiency — to minimize the cost of information processing while maximizing the capacity for growth and adaptation.

Accumulating evidence indicates that general intelligence is associated with global efficiency, the capacity to integrate information across the brain as a whole.

The network neuroscience theory proposes that crystallized intelligence engages highly accessible representations of prior knowledge and experience and relies on easy-to-reach network states.

In contrast, fluid intelligence reflects the capacity to solve novel problems and to demonstrate adaptive, flexible behavior. Fluid intelligence, therefore, engages networks that can transition to difficult-to-reach, highly flexible states.

Thus, rather than attribute intelligence to a fixed set of brain regions or networks, this perspective is based on the dynamic reorganization of brain networks and proposes that intelligence is grounded in brain plasticity."

Read the full study "Investigating cognitive neuroscience theories of human intelligence: A connectome-based predictive modeling approach," published in the journal Human Brain Mapping.

Study abstract:

Central to modern neuroscientific theories of human intelligence is the notion that general intelligence depends on a primary brain region or network, engaging spatially localized (rather than global) neural representations. Recent findings in network neuroscience, however, challenge this assumption, providing evidence that general intelligence may depend on system-wide network mechanisms, suggesting that local representations are necessary but not sufficient to account for the neural architecture of human intelligence. Despite the importance of this key theoretical distinction, prior research has not systematically investigated the role of local versus global neural representations in predicting general intelligence. We conducted a large-scale connectome-based predictive modeling study (N = 297), administering resting-state fMRI and a comprehensive cognitive battery to evaluate the efficacy of modern neuroscientific theories of human intelligence, including spatially localized theories (Lateral Prefrontal Cortex Theory, Parieto-Frontal Integration Theory, and Multiple Demand Theory) and recent global accounts (Process Overlap Theory and Network Neuroscience Theory). The results of our study demonstrate that general intelligence can be predicted by local functional connectivity profiles but is most robustly explained by global profiles of whole-brain connectivity. Our findings further suggest that the improved efficacy of global theories is not reducible to a greater strength or number of connections, but instead results from considering both strong and weak connections that provide the basis for intelligence (as predicted by the Network Neuroscience Theory). Our results highlight the importance of considering local neural representations in the context of a global information-processing architecture, suggesting future directions for theory-driven research on system-wide network mechanisms underlying general intelligence.