A general factor can describe what cognitive tests have in common. But description leaves a harder question untouched: what creates that commonality? Process Overlap Theory begins precisely there. Developed by Kristof Kovacs and Andrew Conway, the theory proposes that different cognitive tasks correlate because they repeatedly recruit overlapping cognitive processes. Complex reasoning, memory and verbal tasks may differ in their specific content while drawing on some of the same domain-general executive processes—maintaining goals, controlling interference, coordinating information and updating intermediate results. On this account, psychometric g remains measurable. What changes is its interpretation. Instead of one psychological resource sitting behind every cognitive task, g can emerge statistically from the overlapping demands of many different tasks.
From a statistical factor to a cognitive mechanism
Return to the central observation. Reasoning, memory, verbal, spatial, and other cognitive tests tend to correlate positively. A common-cause interpretation says:
they correlate because one general ability contributes to all of them.
Process Overlap Theory (POT), developed by Kristof Kovacs and Andrew Conway, asks whether the same covariance could emerge without one single general cognitive process. Its answer is yes.
How overlap can create correlation
Imagine three fictional tasks.
Task A: abstract reasoning
Requires:
- maintaining the current rule;
- resisting a tempting but wrong pattern;
- updating a hypothesis;
- spatial relation processing.
Task B: working-memory challenge
Requires:
- maintaining goals;
- resisting interference;
- updating information;
- short-term retention.
Task C: complex verbal inference
Requires:
- maintaining goals;
- suppressing irrelevant interpretations;
- integrating semantic knowledge;
- relational reasoning.
The tasks are different. But notice the overlap. Goal maintenance appears repeatedly. Interference control appears repeatedly. Updating appears in more than one task. POT proposes that domain-general executive processes are sampled across many cognitive tasks more often than any one domain-specific process. Therefore individual differences in those overlapping processes contribute to correlations across the battery.
So is g fake?
No. That conclusion does not follow. POT does not deny that a general factor can be extracted statistically. It disputes a particular interpretation of that factor. Under a traditional reflective interpretation:
latent g causes the covariance among tests.
Under POT:
overlapping cognitive processes generate the covariance, and
psychometric g summarizes the resulting pattern.
The statistical factor can be real and useful as a summary without being a single causal psychological entity.
Why executive processes?
POT gives special importance to domain-general executive processes associated with maintaining goals and coordinating cognition under interference. The logic is that complex tests often require these processes regardless of their content. A spatial reasoning problem and a verbal reasoning problem may contain very different domain-specific material while both requiring the person to maintain a goal, manage competing representations, and coordinate intermediate steps. That repeated recruitment creates overlap.
Is POT just saying “tests share skills”?
It is more specific than that. A theory becomes useful when it specifies which processes overlap, why the pattern produces the positive manifold, and how the proposal can be formalized and tested. Kovacs and Conway formalized POT using multidimensional measurement models rather than leaving it as a verbal metaphor. But the theory still faces the same standard as its rivals: it must produce evidence that distinguishes its mechanism from other systems capable of generating similar covariance.
What would evidence for POT look like?
Evidence becomes more diagnostic if it shows that:
- different cognitive tasks recruit overlapping executive processes in
the way the theory predicts; - individual differences in those processes account for systematic
covariance; - the pattern of test correlations follows from process overlap rather
than requiring a single common cause; - manipulations or developmental changes affecting overlapping
processes alter covariance in predicted ways.
Simply showing a positive manifold does not uniquely support POT, because common-cause, mutualism, and other models also predict positive correlations.
What is still unresolved?
Several questions remain. Which executive processes are genuinely domain-general? How should they be measured without recreating the same latent-variable problems? How much of the positive manifold can overlap explain? Do developmental and biological influences act on the proposed processes in ways that produce the observed hierarchy? Can POT outperform rival models in genuinely discriminating tests rather than merely fit existing covariance? These are empirical questions.
What Process Overlap changes
POT changes the interpretation of g, not the observation that g can be measured. Its central proposal is:
many tasks overlap in the cognitive processes they require → those
shared process demands create positive correlations → a general factor
emerges statistically.
Whether this is the best causal account remains an active theoretical question.
References
- Kovacs, K., & Conway, A. R. A. (2016). Process Overlap Theory: A
unified account of the general factor of intelligence. Psychological
Inquiry, 27, 151–177. DOI - Kovacs, K., & Conway, A. R. A. (2019). What is IQ? Life beyond
“general intelligence.” Current Directions in Psychological Science,
28. - Schmank, C. J., Goring, S. A., Kovacs, K., & Conway, A. R. A.
(2019). Psychometric network analysis of the Hungarian WAIS. Journal
of Intelligence, 7, 21. - Savi, A. O., et al. (2019). The wiring of intelligence.
Perspectives on Psychological Science, 14, 1034–1061. —. DOI
