Mutualism Theory of Intelligence: Can Cognitive Abilities Make Each Other Stronger?

Most explanations of the positive manifold begin with a pattern that already exists: cognitive abilities are correlated, so what common influence might be producing the correlations? Mutualism asks a more developmental question. What if some of those correlations are not simply present from the beginning? What if they grow because cognitive abilities help one another develop? A child who remembers explanations more effectively may learn more. Greater knowledge can make later reasoning easier. Better reasoning can accelerate subsequent learning. Across years of development, small advantages in different cognitive processes can become connected through repeated reciprocal effects. The mutualism model, introduced by Han van der Maas and colleagues, proposes that interactions of this kind can generate a positive manifold even without assuming that one common cognitive cause drives every ability from the outset.

What is mutualism?

The mutualism model proposed by van der Maas and colleagues begins with a developmental possibility. Imagine early cognitive processes that are not strongly correlated. One child has slightly better memory. Another has slightly better reasoning. Another is more efficient at some perceptual process. During development, those processes interact. Better memory can support learning. Learning produces knowledge that helps reasoning. Better reasoning helps the child understand instruction. Successful learning creates opportunities for still more learning. Positive interactions accumulate. Over time, abilities that began more independently can become positively correlated.

How a developmental loop could form

Consider a fictional child, Sofia. At age five, Sofia is slightly above average at retaining verbal information but average at relational reasoning. Because she retains explanations well, she learns vocabulary and basic numerical concepts efficiently. That accumulated knowledge makes later classroom explanations easier to understand. Better understanding gives her more successful practice with reasoning problems. Those reasoning gains make subsequent learning more efficient. The proposed loop is:

memory advantage → easier learning → more knowledge → easier reasoning → more learning

Now imagine different loops operating across thousands of children. A population-level positive manifold can emerge even if there was no single general cause controlling every ability at the beginning. This is the intuition behind mutualism.

Does mutualism say genes do not matter?

No. Rejecting one common-cause interpretation of g is not equivalent to claiming cognitive development is purely environmental. Genetic differences can influence components of the system, environmental experiences can influence them, and gene–environment correlations and interactions can affect developmental pathways. Mutualism concerns how cognitive processes can become correlated through reciprocal interaction. It is not a nature-versus-nurture theory.

What happens to g in a mutualist account?

Mutualism does not require the statistical general factor to disappear. If developmental interactions make cognitive abilities increasingly correlated, a factor analysis of the resulting test scores can still recover a strong general dimension. The disagreement concerns the direction of explanation. In a simple common-cause picture, a general latent cause produces the correlations among abilities. In mutualism, at least some of the correlations can be an outcome of development: abilities become coupled because gains in one domain alter the conditions for gains in another. That distinction matters because the same cross-sectional covariance pattern can be compatible with very different developmental histories. To test mutualism, researchers therefore need more than a correlation matrix measured at one moment. They need longitudinal evidence about how abilities change together over time.

How do mutualism and Process Overlap differ?

They can sound similar because both reject a simple one-cause reading of g. But their emphases differ. Process Overlap Theory emphasizes overlapping processes recruited by cognitive tasks. Mutualism emphasizes reciprocal developmental interactions among cognitive processes. Wiring/network models emphasize interacting networks of cognitive skills, facts, and procedures and how those networks grow. These mechanisms can potentially coexist.

What would distinguish mutualism from a common-cause model?

This is the difficult scientific question. Both models can generate positive correlations. Evidence becomes more discriminating when it concerns developmental dynamics: whether changes in one cognitive domain prospectively influence changes in another in the reciprocal patterns predicted by mutualism, whether correlation structures change with development as expected, and whether network models predict observations that common-factor models do not. Cross-sectional correlations alone are insufficient.

Mutualism’s wager

Mutualism reverses the usual causal intuition. Instead of saying:

abilities correlate because one general cause sits behind them,

it proposes:

abilities can develop together because they repeatedly influence one another.

The model is scientifically important because it turns the positive manifold into a developmental question: not only what common structure exists? but how did initially differentiated capacities become coupled over time? Its ability to distinguish itself empirically from common-cause and other developmental accounts remains under active investigation.