Process Overlap Theory: Rethinking Intelligence and Cognitive Training

The Mystery of the Positive Manifold

One of the most robust findings in psychology is the positive manifold: cognitive tests tend to correlate with each other. If you're good at vocabulary, you're more likely to be good at spatial reasoning. If you excel at mental arithmetic, you probably do well on reading comprehension too. This observation led to the concept of g, or general intelligence—a single underlying factor that supposedly explains performance across all cognitive domains.

For over a century, g has been treated as a fundamental property of the mind. But what if it's not?

Enter Process Overlap Theory (POT)

Process Overlap Theory offers a radically different explanation: there is no real g or "general intelligence" as a unified cognitive entity. Instead, what we measure as general intelligence in psychometrics is a statistical artifact formed by overlapping cognitive processes that are useful for various cognitive tests.

Think of it this way: imagine you're measuring performance on different sports. You might find that basketball players tend to be decent at volleyball, and volleyball players often do okay at tennis. You could posit a "general athleticism" factor. But what's really happening? These sports share overlapping physical processes—hand-eye coordination, reaction time, spatial awareness, cardiovascular endurance. The correlation isn't mysterious; it's process overlap.

The same logic applies to cognitive tests. They correlate not because of some mystical general intelligence factor, but because they share common cognitive processes.

Why Tests Correlate: The Process Overlap Explanation

Consider these examples:

Raven's Progressive Matrices (often considered a pure measure of fluid intelligence) requires:

• Visual pattern recognition
• Spatial reasoning
• Rule extraction
• Working memory (to hold multiple rules in mind)
• Relational reasoning (understanding how elements relate)

Verbal Comprehension Tests require:

• Vocabulary knowledge (crystallized intelligence)
• Semantic processing
• Working memory (to hold sentence structure)
• Relational reasoning (understanding how words and concepts relate)

Mental Arithmetic requires:

• Numerical processing
• Working memory (to hold intermediate results)
• Sequential processing
• Automated math facts (crystallized knowledge)

Notice the overlap? Working memory appears in all three. Relational reasoning appears in at least two. This overlap is why these tests correlate—not because of g, but because they tap into shared cognitive processes.

The Problem with Isolating Cognitive Processes

POT reveals a fundamental challenge in cognitive testing: it's incredibly difficult to completely isolate a single cognitive process. Nearly every test requires multiple processes working together.

Even tests that seem highly specific still correlate with g. Take vocabulary tests—they're about word knowledge, right? Yet they correlate with measures of fluid intelligence. How can that be explained without invoking general intelligence?

The Investment Theory Connection

This is where the Investment Theory of Intelligence comes in. It proposes that crystallized intelligence (accumulated knowledge like vocabulary) is heavily influenced by fluid intelligence (reasoning ability) over a lifetime.

Here's the mechanism:

1. People with better fluid intelligence find learning easier and more rewarding
2. Because learning is easier, they engage in it more often
3. Because they're successful, learning becomes more motivating
4. Over years and decades, they accumulate more knowledge

So vocabulary doesn't correlate with IQ because of some general intelligence factor—it correlates because people who reason better have spent a lifetime learning more words more efficiently. The correlation is explained by process overlap (the reasoning processes used in learning) and developmental investment, not by g.

Why This Matters for Cognitive Training

POT has profound implications for how we should approach cognitive training. It explains two major patterns:

Why Task-Specific Training Fails

Training on highly specific tasks—like the Wisconsin Card Sorting Test, n-back tasks, or even chess—rarely transfers to other domains. POT explains why:

When you train a specific complex task, you're training:

• Shared processes (working memory, attention, etc.)
• Task-specific strategies (unique to that particular test)
• Task-specific knowledge (the particular patterns in that task)

The problem? It's easier for the brain to get better at the task-specific elements than to strengthen the underlying shared processes. You end up with someone who's excellent at n-back but whose working memory in other contexts hasn't improved.

It's like a pianist practicing only 'Flight of the Bumblebee' at lightning speed and expecting to become a better all-around musician. They'll master that one incredibly difficult piece (task-specific knowledge), but it won't improve their ability to improvise jazz, compose a melody, or sensitively accompany a singer. They've trained a specific technical feat, not the foundational processes of musicality like understanding harmony, rhythm, and theory.

Why Foundational Process Training Works

But what if you could train the shared processes directly? This is where POT points toward a solution:

Train processes that are:

1. Foundational (used across many cognitive tasks)
2. General (not tied to specific content or strategies)
3. Abstract (stripped of task-specific elements)

This is exactly what programs like SMART do. By training relational reasoning using abstract stimuli:

• You can't develop task-specific strategies (there's no content to memorize)
• You're forced to engage the pure process
• The process is so foundational that it's used in language, logic, math, reading, and countless other domains

When you strengthen relational reasoning, you get genuine near transfer to every task that uses relational reasoning—which, according to POT, is most cognitive tasks. It's not "far transfer" in the magical sense; it's near transfer to a foundational process that happens to be used everywhere.

The POT Prescription for Cognitive Training

Based on Process Overlap Theory, effective cognitive training should target:

1. Relational Reasoning

The ability to understand and derive relationships between concepts. This is perhaps the most fundamental cognitive process, underlying language, logic, mathematics, and complex thought.

2. Working Memory

The ability to hold and manipulate information in mind. This appears as a component in virtually every cognitive test and real-world cognitive task.

3. Processing Speed

How quickly you can perceive and respond to information. This is a fundamental bottleneck for many cognitive operations.

4. Verbal Fluency

The efficiency of lexical access and verbal processing. Given how much of human thought is verbal, this is a core process with broad impact.

Notice something? These are exactly the four modules in Relatoria:

• Reason trains relational reasoning (via SMART and advanced protocols)
• Remember trains working memory (18+ specialized exercises)
• Fluent trains verbal fluency (multiple training modes)
• Process trains processing speed (UFOV-inspired training)

POT vs. g: What's the Practical Difference?

You might think: "If tests correlate anyway, does it matter whether we call it g or process overlap?"

It matters enormously for training strategy:

If g exists as a unified entity:

• We'd need to find a way to train "general intelligence" directly
• Training specific processes shouldn't work
• Transfer should be all-or-nothing

If POT is correct:

• We should train the most foundational, widely-used processes
• We should expect near transfer to tasks using those processes
• Transfer is predictable based on process overlap
• Abstract, general training should outperform content-specific training

The evidence supports POT. Task-specific training fails. Training foundational processes with abstract stimuli (like SMART) succeeds.

Cognitive Architecture: Where POT Meets Neuroscience

Some might wonder: if there's no g, why do brain imaging studies find correlations between intelligence and certain brain networks?

This is where POT is compatible with models like P-FIT (Parieto-Frontal Integration Theory). P-FIT shows that certain brain regions—particularly parietal and frontal areas—are associated with intelligence measures.

But POT would interpret this differently: these aren't the seat of "general intelligence." These are the neural implementations of the foundational cognitive processes. The parieto-frontal network handles working memory, attention, and relational reasoning—exactly the processes that appear across many cognitive tests.

P-FIT shows where the processes are in the brain. POT explains why they correlate. They're complementary, not contradictory.

Functional Cognitive Training: The POT-Informed Approach

Understanding POT leads to a principle: cognitive training tasks must be functionally similar to useful cognitive requirements while training foundational processes.

This means:

• Abstract enough to avoid task-specific strategies
• General enough to engage pure processes
• Similar enough to real cognitive demands to ensure the trained process is actually useful

For example:

• Training verbal span (remembering sequences of words) is better than training digit span, because verbal working memory is more functionally similar to real-world thinking
• Training relational reasoning with abstract stimuli is better than training with specific content domains, because it engages the pure process
• Training with mastery requirements ensures the process becomes automatic, not just consciously accessible

The Bottom Line: Near Transfer Is All There Is

Time and time again, research shows that only near transfer exists. POT explains this perfectly: you improve in tasks that use the processes you trained.

What researchers sometimes call "far transfer" is really just near transfer to a foundational process that happens to be useful in many domains. It seems "far" because the tasks look different, but it's "near" at the process level.

This is liberating. We don't need to achieve magical far transfer. We just need to:

1. Identify the most foundational cognitive processes
2. Train them in abstract, general ways
3. Demand mastery so they become automatic
4. Reap the benefits across all tasks using those processes

Conclusion: Training Smart, Not General

Process Overlap Theory transforms how we think about intelligence and training. There's no mystical g to chase. There are foundational cognitive processes to strengthen.

By training relational reasoning, working memory, verbal fluency, and processing speed—the core overlapping processes underlying most cognitive tasks—we can achieve real, lasting cognitive improvement that transfers broadly because the trained processes are used broadly.

This isn't about becoming "generally smarter" in some vague sense. It's about strengthening the specific cognitive engines that power thinking across domains. That's the Process Overlap Theory approach, and that's why it works.

Ready to train your foundational processes? Start with SMART—completely free.